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-itaiji-016
 377     =+>ucs-var-003
 378     =>ucs-itaiji-001
 379     =>ucs-itaiji-002
 380     =>ucs-itaiji-003
 381     =>ucs-itaiji-004
 382     =>ucs-itaiji-005
 383     =>ucs-itaiji-006
 384     =>ucs-itaiji-007
 385     =>ucs-itaiji-008
 386     =>ucs-itaiji-009
 387     =>ucs-itaiji-010
 388     =>ucs-itaiji-011
 389     =big5-cdp-var-001
 390     =big5-cdp-var-002
 391     =big5-cdp-var-003
 392     =big5-cdp-var-004
 393     =big5-cdp-var-005
 394     =big5-cdp-var-007
 395     =big5-cdp-var-010
 396     =big5-cdp-itaiji-001
 397     =big5-cdp-itaiji-002
 398     =big5-cdp-itaiji-003
 399     =big5-cdp-itaiji-004
 400     =big5-cdp-itaiji-005
 401     =>big5-cdp-itaiji-001
 402     =>big5-cdp-itaiji-002
 403     =>big5-cdp-itaiji-003
 404     =>big5-cdp-itaiji-004
 405     =>big5-cdp-var-001
 406     =>ucs@iwds-1
 407     =>ucs@cognate
 408     =>ucs@component
 409     =>ucs@iwds-1/normalized
 410     =>ucs-itaiji-001@iwds-1
 411     =>ucs-itaiji-002@iwds-1
 412     =>ucs-itaiji-003@iwds-1
 413     =>ucs-itaiji-006@iwds-1
 414     =>big5-cdp@component
 415     =>big5-cdp@cognate
 416     ==ucs@gb
 417     =ucs@gb
 418     ==ucs-var-002
 419     =ucs@JP/hanazono
 420     =koseki
 421     =shuowen-jiguge
 422     ===shuowen-jiguge5
 423     ===shuowen-jiguge4))
 424
 425
 426 ;;; @ char-db formatters
 427 ;;;
 428
 429 (defun char-db-make-char-spec (char)
 430   (let (ret char-spec)
 431     (cond ((characterp char)
 432            (cond ((and (setq ret (encode-char char '=ucs 'defined-only))
 433                        (not (and (<= #xE000 ret)(<= ret #xF8FF))))
 434                   (setq char-spec (list (cons '=ucs ret)))
 435                   (cond ((setq ret (get-char-attribute char 'name))
 436                          (setq char-spec (cons (cons 'name ret) char-spec))
 437                          )
 438                         ((setq ret (get-char-attribute char 'name*))
 439                          (setq char-spec (cons (cons 'name* ret) char-spec))
 440                          ))
 441                   )
 442                  ((encode-char char '=mj 'defined-only)
 443                   (setq char-spec nil)
 444                   (dolist (ccs (charset-list))
 445                     (if (and (or (eq ccs '=mj)
 446                                  ;; (eq (charset-property ccs 'iso-ir) 177)
 447                                  (string-match "=ucs@" (symbol-name ccs))
 448                                  )
 449                              (setq ccs (charset-name ccs))
 450                              (null (assq ccs char-spec))
 451                              (setq ret (encode-char char ccs 'defined-only)))
 452                         (setq char-spec (cons (cons ccs ret) char-spec))))
 453                   )
 454                  ((encode-char char '==mj 'defined-only)
 455                   (setq char-spec nil)
 456                   (dolist (ccs (charset-list))
 457                     (if (and (or (eq ccs '==mj)
 458                                  ;; (eq (charset-property ccs 'iso-ir) 177)
 459                                  (string-match "=ucs@" (symbol-name ccs))
 460                                  )
 461                              (setq ccs (charset-name ccs))
 462                              (null (assq ccs char-spec))
 463                              (setq ret (encode-char char ccs 'defined-only)))
 464                         (setq char-spec (cons (cons ccs ret) char-spec))))
 465                   )
 466                  ((encode-char char '=adobe-japan1 'defined-only)
 467                   (setq char-spec nil)
 468                   (dolist (ccs (charset-list))
 469                     (if (and (or (memq ccs
 470                                        '(=adobe-japan1-0
 471                                          =adobe-japan1-1
 472                                          =adobe-japan1-2
 473                                          =adobe-japan1-3
 474                                          =adobe-japan1-4
 475                                          =adobe-japan1-5
 476                                          =adobe-japan1-6
 477                                          ))
 478                                  ;; (eq (charset-property ccs 'iso-ir) 177)
 479                                  (string-match "=ucs@" (symbol-name ccs))
 480                                  )
 481                              (setq ccs (charset-name ccs))
 482                              (null (assq ccs char-spec))
 483                              (setq ret (encode-char char ccs 'defined-only)))
 484                         (setq char-spec (cons (cons ccs ret) char-spec))))
 485                   )
 486                  ((encode-char char '==adobe-japan1 'defined-only)
 487                   (setq char-spec nil)
 488                   (dolist (ccs (charset-list))
 489                     (if (and (or (memq ccs
 490                                        '(==adobe-japan1-0
 491                                          ==adobe-japan1-1
 492                                          ==adobe-japan1-2
 493                                          ==adobe-japan1-3
 494                                          ==adobe-japan1-4
 495                                          ==adobe-japan1-5
 496                                          ==adobe-japan1-6
 497                                          ))
 498                                  ;; (eq (charset-property ccs 'iso-ir) 177)
 499                                  (string-match "=ucs@" (symbol-name ccs))
 500                                  )
 501                              (setq ccs (charset-name ccs))
 502                              (null (assq ccs char-spec))
 503                              (setq ret (encode-char char ccs 'defined-only)))
 504                         (setq char-spec (cons (cons ccs ret) char-spec))))
 505                   )
 506                  ((setq ret
 507                         (catch 'tag
 508                           (let ((rest char-db-coded-charset-priority-list)
 509                                 ccs)
 510                             (while rest
 511                               (setq ccs (charset-name
 512                                          (find-charset (car rest))))
 513                               (if (setq ret
 514                                         (encode-char char ccs
 515                                                      'defined-only))
 516                                   (throw 'tag (cons ccs ret)))
 517                               (setq rest (cdr rest))))))
 518                   (setq char-spec (list ret))
 519                   (dolist (ccs (delq (car ret) (charset-list)))
 520                     (if (and (or (charset-iso-final-char ccs)
 521                                  (memq ccs
 522                                        '(=daikanwa
 523                                          =daikanwa@rev2
 524                                          ;; =gt-k
 525                                          =jis-x0208@1997
 526                                          ))
 527                                  (and (eq (charset-property ccs 'iso-ir) 177)
 528                                       (not (eq (charset-name ccs) '=ucs@big5)))
 529                                  ;; (string-match "=ucs@" (symbol-name ccs))
 530                                  )
 531                              (setq ccs (charset-name ccs))
 532                              (null (assq ccs char-spec))
 533                              (setq ret (encode-char char ccs 'defined-only)))
 534                         (setq char-spec (cons (cons ccs ret) char-spec))))
 535                   (if (null char-spec)
 536                       (setq char-spec (split-char char)))
 537                   (cond ((setq ret (get-char-attribute char 'name))
 538                          (setq char-spec (cons (cons 'name ret) char-spec))
 539                          )
 540                         ((setq ret (get-char-attribute char 'name*))
 541                          (setq char-spec (cons (cons 'name* ret) char-spec))
 542                          ))
 543                   )
 544                  ((setq ret (get-char-attribute
 545                              char 'ideographic-combination))
 546                   (setq char-spec
 547                         (cons (cons 'ideographic-combination ret)
 548                               char-spec))
 549                   ))
 550            char-spec)
 551           ((consp char)
 552            char))))
 553
 554 (defun char-db-insert-char-spec (char &optional readable column
 555                                       required-features)
 556   (unless column
 557     (setq column (current-column)))
 558   (let (char-spec temp-char)
 559     (setq char-spec (char-db-make-char-spec char))
 560     (unless (or (characterp char) ; char
 561                 (condition-case nil
 562                     (setq char (find-char char-spec))
 563                   (error nil)))
 564       ;; define temporary character
 565       ;;   Current implementation is dirty.
 566       (setq temp-char (define-char (cons '(ideograph-daikanwa . 0)
 567                                          char-spec)))
 568       (remove-char-attribute temp-char 'ideograph-daikanwa)
 569       (setq char temp-char))
 570     (insert-char-attributes char
 571                             readable
 572                             (union (mapcar #'car char-spec)
 573                                    required-features)
 574                             nil 'for-sub-node)
 575     (when temp-char
 576       ;; undefine temporary character
 577       ;;   Current implementation is dirty.
 578       (setq char-spec (char-attribute-alist temp-char))
 579       (while char-spec
 580         (remove-char-attribute temp-char (car (car char-spec)))
 581         (setq char-spec (cdr char-spec))))))
 582
 583 (defun char-db-insert-alist (alist &optional readable column)
 584   (unless column
 585     (setq column (current-column)))
 586   (let ((line-breaking
 587          (concat "\n" (make-string (1+ column) ?\ )))
 588         name value
 589         ret al ; cal
 590         key
 591         lbs cell rest separator)
 592     (insert "(")
 593     (while alist
 594       (setq name (car (car alist))
 595             value (cdr (car alist)))
 596       (cond ((eq name 'char)
 597              (insert "(char . ")
 598              (if (setq ret (condition-case nil
 599                                (find-char value)
 600                              (error nil)))
 601                  (progn
 602                    (setq al nil
 603                          ;; cal nil
 604                          )
 605                    (while value
 606                      (setq key (car (car value)))
 607                      ;; (if (find-charset key)
 608                      ;;     (setq cal (cons key cal))
 609                      (setq al (cons key al))
 610                      ;; )
 611                      (setq value (cdr value)))
 612                    (insert-char-attributes ret
 613                                            readable
 614                                            (or al 'none) ; cal
 615                                            nil 'for-sub-node))
 616                (insert (prin1-to-string value)))
 617              (insert ")")
 618              (insert line-breaking))
 619             ((consp value)
 620              (insert (format "(%-18s " name))
 621              (setq lbs (concat "\n" (make-string (current-column) ?\ )))
 622              (while (consp value)
 623                (setq cell (car value))
 624                (if (and (consp cell)
 625                         (consp (car cell))
 626                         (setq ret (condition-case nil
 627                                       (find-char cell)
 628                                     (error nil)))
 629                         )
 630                    (progn
 631                      (setq rest cell
 632                            al nil
 633                            ;; cal nil
 634                            )
 635                      (while rest
 636                        (setq key (car (car rest)))
 637                        ;; (if (find-charset key)
 638                        ;;     (setq cal (cons key cal))
 639                        (setq al (cons key al))
 640                        ;; )
 641                        (setq rest (cdr rest)))
 642                      (if separator
 643                          (insert lbs))
 644                      (insert-char-attributes ret
 645                                              readable
 646                                              al ; cal
 647                                              nil 'for-sub-node)
 648                      (setq separator lbs))
 649                  (if separator
 650                      (insert separator))
 651                  (insert (prin1-to-string cell))
 652                  (setq separator " "))
 653                (setq value (cdr value)))
 654              (insert ")")
 655              (insert line-breaking))
 656             (t
 657              (insert (format "(%-18s . %S)%s"
 658                              name value
 659                              line-breaking))))
 660       (setq alist (cdr alist))))
 661   (insert ")"))
 662
 663 (defun char-db-insert-char-reference (plist &optional readable column)
 664   (unless column
 665     (setq column (current-column)))
 666   (let ((line-breaking
 667          (concat "\n" (make-string (1+ column) ?\ )))
 668         (separator "")
 669         name value)
 670     (insert "(")
 671     (while plist
 672       (setq name (pop plist))
 673       (setq value (pop plist))
 674       (cond ((eq name :char)
 675              (insert separator)
 676              (insert ":char\t")
 677              (cond ((numberp value)
 678                     (setq value (decode-char '=ucs value)))
 679                    ;; ((consp value)
 680                    ;;  (setq value (or (find-char value)
 681                    ;;                  value)))
 682                    )
 683              (char-db-insert-char-spec value readable)
 684              (insert line-breaking)
 685              (setq separator ""))
 686             ((eq name :radical)
 687              (insert (format "%s%s\t%d ; %c%s"
 688                              separator
 689                              name value
 690                              (ideographic-radical value)
 691                              line-breaking))
 692              (setq separator ""))
 693             (t
 694              (insert (format "%s%s\t%S" separator name value))
 695              (setq separator line-breaking)))
 696       ))
 697   (insert ")"))
 698
 699 (defun char-db-decode-isolated-char (ccs code-point)
 700   (let (ret)
 701     (setq ret
 702           (cond ((eq ccs 'arabic-iso8859-6)
 703                  (decode-char ccs code-point))
 704                 ;; ((eq ccs '=gt)
 705                 ;;  (decode-builtin-char '==gt code-point))
 706                 ((and (memq ccs '(=gt-pj-1
 707                                   =gt-pj-2
 708                                   =gt-pj-3
 709                                   =gt-pj-4
 710                                   =gt-pj-5
 711                                   =gt-pj-6
 712                                   =gt-pj-7
 713                                   =gt-pj-8
 714                                   =gt-pj-9
 715                                   =gt-pj-10
 716                                   =gt-pj-11))
 717                       (setq ret (decode-char ccs code-point))
 718                       (setq ret (encode-char ret '=gt 'defined-only)))
 719                  (decode-builtin-char '=gt ret))
 720                 (t
 721                  (decode-builtin-char ccs code-point))))
 722     (cond ((null ret)
 723            (or (decode-char ccs code-point)
 724                (define-char (list (cons ccs code-point)))))
 725           ((and (<= 0 (char-int ret))
 726                 (<= (char-int ret) #x1F))
 727            (decode-char '=ucs (+ #x2400 (char-int ret))))
 728           ((= (char-int ret) #x7F)
 729            ?\u2421)
 730           (t ret))))
 731
 732 (defvar char-db-convert-obsolete-format t)
 733
 734 (defun char-db-insert-ccs-feature (name value line-breaking)
 735   (cond
 736    ((integerp value)
 737     (insert
 738      (format
 739       (cond
 740        ((memq name '(=>iwds-1
 741                      ==shinjigen
 742                      =shinjigen
 743                      =shinjigen@1ed ==shinjigen@1ed
 744                      =shinjigen@rev ==shinjigen@rev
 745                      =shinjigen/+p@rev ==shinjigen/+p@rev
 746                      ===daikanwa/ho ==daikanwa/ho
 747                      =daikanwa/ho =>>daikanwa/ho =>daikanwa/ho))
 748         "(%-18s .  %04d)\t; %c")
 749        ((eq name '=shinjigen@1ed/24pr)
 750         "(%-18s . %04d)\t; %c")
 751        ((or
 752          (memq name
 753                '(===daikanwa
 754                  ==daikanwa =daikanwa =>>daikanwa =>daikanwa
 755                  =daikanwa@rev1 =daikanwa@rev2
 756                  =daikanwa/+p ==daikanwa/+p ===daikanwa/+p
 757                  =>>daikanwa/+p
 758                  =daikanwa/+2p =>>daikanwa/+2p
 759                  =gt ==gt ===gt
 760                  =>>gt =+>gt =>gt
 761                  =gt-k ==gt-k ===gt-k
 762                  =>>gt-k =>gt-k
 763                  =adobe-japan1-0 ==adobe-japan1-0 ===adobe-japan1-0
 764                  =adobe-japan1-1 ==adobe-japan1-1 ===adobe-japan1-1
 765                  =adobe-japan1-2 ==adobe-japan1-2 ===adobe-japan1-2
 766                  =adobe-japan1-3 ==adobe-japan1-3 ===adobe-japan1-3
 767                  =adobe-japan1-4 ==adobe-japan1-4 ===adobe-japan1-4
 768                  =adobe-japan1-5 ==adobe-japan1-5 ===adobe-japan1-5
 769                  =adobe-japan1-6 ==adobe-japan1-6 ===adobe-japan1-6
 770                  =>>adobe-japan1-0 =+>adobe-japan1-0
 771                  =>>adobe-japan1-1 =+>adobe-japan1-1
 772                  =>>adobe-japan1-2 =+>adobe-japan1-2
 773                  =>>adobe-japan1-3 =+>adobe-japan1-3
 774                  =>>adobe-japan1-4 =+>adobe-japan1-4
 775                  =>>adobe-japan1-5 =+>adobe-japan1-5
 776                  =>>adobe-japan1-6 =+>adobe-japan1-6
 777                  =>cbeta =cbeta =>>cbeta ==cbeta ===cbeta
 778                  =zinbun-oracle =>zinbun-oracle
 779                  =daijiten ==daijiten ===daijiten ==>daijiten
 780                  ===hng-jou ===hng-keg ===hng-dng ===hng-mam
 781                  ===hng-drt ===hng-kgk ===hng-myz ===hng-kda
 782                  ===hng-khi ===hng-khm ===hng-fhs ===hng-hok
 783                  ===hng-kyd ===hng-sok
 784                  ===hng-yhk ===hng-kak ===hng-kar ===hng-kae
 785                  ===hng-sys ===hng-tsu ===hng-tzj
 786                  ===hng-hos ===hng-kkh ===hng-nak ===hng-jhk
 787                  ===hng-hod ===hng-gok ===hng-ink ===hng-nto
 788                  ===hng-nkm ===hng-k24 ===hng-ini ===hng-nkk
 789                  ===hng-kcc ===hng-kcj ===hng-kbk ===hng-sik
 790                  ===hng-skk ===hng-kyu ===hng-ksk ===hng-wan
 791                  ===hng-okd ===hng-wad ===hng-kmi
 792                  ===hng-zkd ===hng-doh ===hng-jyu ===hng-tzs
 793                  ===hng-sai ===hng-kad ===hng-kss
 794                  ===hng-kyo ===hng-ykk ===hng-saa
 795                  ===hng-sab ===hng-wks ===hng-wke ===hng-smk
 796                  ===hng-sgs ===hng-sts ===hng-kkd
 797                  =shuowen-jiguge ===shuowen-jiguge4 ===shuowen-jiguge5))
 798          ;; (string-match "^=adobe-" (symbol-name name))
 799          )
 800         "(%-18s . %05d)\t; %c")
 801        ((memq name '(=hanyo-denshi/ks
 802                      ==hanyo-denshi/ks ===hanyo-denshi/ks
 803                      =>>hanyo-denshi/ks
 804                      =koseki ==koseki
 805                      =mj ==mj ===mj =>>mj =>mj
 806                      =zihai mojikyo))
 807         "(%-18s . %06d)\t; %c")
 808        ((memq name '(=hanyo-denshi/tk ==hanyo-denshi/tk))
 809         "(%-18s . %08d)\t; %c")
 810        ((>= (charset-dimension name) 2)
 811         "(%-18s . #x%04X)\t; %c")
 812        (t
 813         "(%-18s . #x%02X)\t; %c"))
 814       name
 815       (if (= (charset-iso-graphic-plane name) 1)
 816           (logior value
 817                   (cond ((= (charset-dimension name) 1)
 818                          #x80)
 819                         ((= (charset-dimension name) 2)
 820                          #x8080)
 821                         ((= (charset-dimension name) 3)
 822                          #x808080)
 823                         (t 0)))
 824         value)
 825       (char-db-decode-isolated-char name value)))
 826     (if (and (= (charset-chars name) 94)
 827              (= (charset-dimension name) 2))
 828         (insert (format " [%02d-%02d]"
 829                         (- (lsh value -8) 32)
 830                         (- (logand value 255) 32))))
 831     )
 832    (t
 833     (insert (format "(%-18s . %s)" name value))
 834     ))
 835   (insert line-breaking))
 836
 837 (defun char-db-insert-relation-feature (char name value line-breaking
 838                                              ccss readable)
 839   (insert (format "(%-18s%s " name line-breaking))
 840   (let ((lbs (concat "\n" (make-string (current-column) ?\ )))
 841         separator cell sources required-features
 842         ret)
 843     (while (consp value)
 844       (setq cell (car value))
 845       (if (integerp cell)
 846           (setq cell (decode-char '=ucs cell)))
 847       (cond
 848        ((eq name '->subsumptive)
 849         (when (or (not (some (lambda (atr)
 850                                (get-char-attribute cell atr))
 851                              char-db-ignored-attributes))
 852                   (some (lambda (ccs)
 853                           (encode-char cell ccs 'defined-only))
 854                         ccss))
 855           (if separator
 856               (insert lbs))
 857           (let ((char-db-ignored-attributes
 858                  (cons '<-subsumptive
 859                        char-db-ignored-attributes)))
 860             (insert-char-attributes cell readable nil nil 'for-sub-node))
 861           (setq separator lbs))
 862         )
 863        ((characterp cell)
 864         (setq sources
 865               (get-char-attribute
 866                char (intern (format "%s*sources" name))))
 867         (setq required-features nil)
 868         (dolist (source sources)
 869           (cond
 870            ((memq source '(JP
 871                            JP/Jouyou
 872                            shinjigen shinjigen@1ed shinjigen@rev))
 873             (setq required-features
 874                   (union required-features
 875                          '(=jis-x0208
 876                            =jis-x0208@1990
 877                            =jis-x0213-1@2000
 878                            =jis-x0213-1@2004
 879                            =jis-x0213-2
 880                            =jis-x0212
 881                            =jis-x0208@1983
 882                            =jis-x0208@1978
 883                            =shinjigen
 884                            =shinjigen@1ed
 885                            =shinjigen@rev
 886                            =shinjigen/+p@rev))))
 887            ((eq source 'CN)
 888             (setq required-features
 889                   (union required-features
 890                          '(=gb2312
 891                            =gb12345
 892                            =iso-ir165)))))
 893           (cond
 894            ((find-charset (setq ret (intern (format "=%s" source))))
 895             (setq required-features
 896                   (cons ret required-features)))
 897            (t (setq required-features
 898                     (cons source required-features)))))
 899         (cond ((string-match "@JP" (symbol-name name))
 900                (setq required-features
 901                      (union required-features
 902                             '(=jis-x0208
 903                               =jis-x0208@1990
 904                               =jis-x0213-1-2000
 905                               =jis-x0213-2-2000
 906                               =jis-x0212
 907                               =jis-x0208@1983
 908                               =jis-x0208@1978))))
 909               ((string-match "@CN" (symbol-name name))
 910                (setq required-features
 911                      (union required-features
 912                             '(=gb2312
 913                               =gb12345
 914                               =iso-ir165)))))
 915         (if separator
 916             (insert lbs))
 917         (if readable
 918             (insert (format "%S" cell))
 919           (char-db-insert-char-spec cell readable
 920                                     nil
 921                                     required-features))
 922         (setq separator lbs))
 923        ((consp cell)
 924         (if separator
 925             (insert lbs))
 926         (if (consp (car cell))
 927             (char-db-insert-char-spec cell readable)
 928           (char-db-insert-char-reference cell readable))
 929         (setq separator lbs))
 930        (t
 931         (if separator
 932             (insert separator))
 933         (insert (prin1-to-string cell))
 934         (setq separator " ")))
 935       (setq value (cdr value)))
 936     (insert ")")
 937     (insert line-breaking)))
 938
 939 (defun insert-char-attributes (char &optional readable attributes column
 940                                     for-sub-node)
 941   (unless column
 942     (setq column (current-column)))
 943   (let (name value ; has-long-ccs-name
 944         rest
 945         radical strokes
 946         (line-breaking
 947          (concat "\n" (make-string (1+ column) ?\ )))
 948         lbs cell separator ret
 949         key al cal
 950         dest-ccss ; sources required-features
 951         ccss)
 952     (let (atr-d)
 953       (setq attributes
 954             (sort (if attributes
 955                       (if (consp attributes)
 956                           (progn
 957                             (dolist (name attributes)
 958                               (unless (memq name char-db-ignored-attributes)
 959                                 (if (find-charset name)
 960                                     (push name ccss))
 961                                 (push name atr-d)))
 962                             atr-d))
 963                     (dolist (name (char-attribute-list))
 964                       (unless (memq name char-db-ignored-attributes)
 965                         (if (find-charset name)
 966                             (push name ccss))
 967                         (push name atr-d)))
 968                     atr-d)
 969                   #'char-attribute-name<)))
 970     (insert "(")
 971     (when (memq '<-subsumptive attributes)
 972       (when (or readable (not for-sub-node))
 973         (when (setq value (get-char-attribute char '<-subsumptive))
 974           (char-db-insert-relation-feature char '<-subsumptive value
 975                                            line-breaking
 976                                            ccss readable)))
 977       (setq attributes (delq '<-subsumptive attributes)))
 978     (when (and (memq '<-denotational attributes)
 979                (setq value (get-char-attribute char '<-denotational)))
 980       (char-db-insert-relation-feature char '<-denotational value
 981                                        line-breaking
 982                                        ccss readable)
 983       (setq attributes (delq '<-denotational attributes)))
 984     (when (and (memq '<-denotational@component attributes)
 985                (setq value (get-char-attribute char '<-denotational@component)))
 986       (char-db-insert-relation-feature char '<-denotational@component value
 987                                        line-breaking
 988                                        ccss readable)
 989       (setq attributes (delq '<-denotational@component attributes)))
 990     (when (and (memq '<-denotational@usage attributes)
 991                (setq value (get-char-attribute char '<-denotational@usage)))
 992       (char-db-insert-relation-feature char '<-denotational@usage value
 993                                        line-breaking
 994                                        ccss readable)
 995       (setq attributes (delq '<-denotational@usage attributes)))
 996     (when (and (memq 'name attributes)
 997                (setq value (get-char-attribute char 'name)))
 998       (insert (format
 999                (if (> (+ (current-column) (length value)) 48)
1000                    "(name . %S)%s"
1001                  "(name               . %S)%s")
1002                value line-breaking))
1003       (setq attributes (delq 'name attributes))
1004       )
1005     (when (and (memq 'name* attributes)
1006                (setq value (get-char-attribute char 'name*)))
1007       (insert (format
1008                (if (> (+ (current-column) (length value)) 48)
1009                    "(name* . %S)%s"
1010                  "(name*              . %S)%s")
1011                value line-breaking))
1012       (setq attributes (delq 'name* attributes))
1013       )
1014     (when (and (memq 'script attributes)
1015                (setq value (get-char-attribute char 'script)))
1016       (insert (format "(script\t\t%s)%s"
1017                       (mapconcat (function prin1-to-string)
1018                                  value " ")
1019                       line-breaking))
1020       (setq attributes (delq 'script attributes))
1021       )
1022     (dolist (name '(=>ucs =>ucs*))
1023       (when (and (memq name attributes)
1024                  (setq value (get-char-attribute char name)))
1025         (insert (format "(%-18s . #x%04X)\t; %c%s"
1026                         name value (decode-char '=ucs value)
1027                         line-breaking))
1028         (setq attributes (delq name attributes))))
1029     (dolist (name '(=>ucs@gb =>ucs@big5))
1030       (when (and (memq name attributes)
1031                  (setq value (get-char-attribute char name)))
1032         (insert (format "(%-18s . #x%04X)\t; %c%s"
1033                         name value
1034                         (decode-char (intern
1035                                       (concat "="
1036                                               (substring
1037                                                (symbol-name name) 2)))
1038                                      value)
1039                         line-breaking))
1040         (setq attributes (delq name attributes))
1041         ))
1042     ;; (dolist (name '(=>daikanwa))
1043     ;;   (when (and (memq name attributes)
1044     ;;              (setq value (get-char-attribute char name)))
1045     ;;     (insert
1046     ;;      (if (integerp value)
1047     ;;          (format "(%-18s . %05d)\t; %c%s"
1048     ;;                  name value (decode-char '=daikanwa value)
1049     ;;                  line-breaking)
1050     ;;        (format "(%-18s %s)\t; %c%s"
1051     ;;                name
1052     ;;                (mapconcat (function prin1-to-string)
1053     ;;                           value " ")
1054     ;;                (char-representative-of-daikanwa char)
1055     ;;                line-breaking)))
1056     ;;     (setq attributes (delq name attributes))))
1057     (when (and (memq 'general-category attributes)
1058                (setq value (get-char-attribute char 'general-category)))
1059       (insert (format
1060                "(general-category\t%s) ; %s%s"
1061                (mapconcat (lambda (cell)
1062                             (format "%S" cell))
1063                           value " ")
1064                (cond ((rassoc value unidata-normative-category-alist)
1065                       "Normative Category")
1066                      ((rassoc value unidata-informative-category-alist)
1067                       "Informative Category")
1068                      (t
1069                       "Unknown Category"))
1070                line-breaking))
1071       (setq attributes (delq 'general-category attributes))
1072       )
1073     (when (and (memq 'bidi-category attributes)
1074                (setq value (get-char-attribute char 'bidi-category)))
1075       (insert (format "(bidi-category\t. %S)%s"
1076                       value
1077                       line-breaking))
1078       (setq attributes (delq 'bidi-category attributes))
1079       )
1080     (unless (or (not (memq 'mirrored attributes))
1081                 (eq (setq value (get-char-attribute char 'mirrored 'empty))
1082                     'empty))
1083       (insert (format "(mirrored\t\t. %S)%s"
1084                       value
1085                       line-breaking))
1086       (setq attributes (delq 'mirrored attributes))
1087       )
1088     (cond
1089      ((and (memq 'decimal-digit-value attributes)
1090            (setq value (get-char-attribute char 'decimal-digit-value)))
1091       (insert (format "(decimal-digit-value . %S)%s"
1092                       value
1093                       line-breaking))
1094       (setq attributes (delq 'decimal-digit-value attributes))
1095       (when (and (memq 'digit-value attributes)
1096                  (setq value (get-char-attribute char 'digit-value)))
1097         (insert (format "(digit-value\t . %S)%s"
1098                         value
1099                         line-breaking))
1100         (setq attributes (delq 'digit-value attributes))
1101         )
1102       (when (and (memq 'numeric-value attributes)
1103                  (setq value (get-char-attribute char 'numeric-value)))
1104         (insert (format "(numeric-value\t . %S)%s"
1105                         value
1106                         line-breaking))
1107         (setq attributes (delq 'numeric-value attributes))
1108         )
1109       )
1110      (t
1111       (when (and (memq 'digit-value attributes)
1112                  (setq value (get-char-attribute char 'digit-value)))
1113         (insert (format "(digit-value\t. %S)%s"
1114                         value
1115                         line-breaking))
1116         (setq attributes (delq 'digit-value attributes))
1117         )
1118       (when (and (memq 'numeric-value attributes)
1119                  (setq value (get-char-attribute char 'numeric-value)))
1120         (insert (format "(numeric-value\t. %S)%s"
1121                         value
1122                         line-breaking))
1123         (setq attributes (delq 'numeric-value attributes))
1124         )))
1125     (when (and (memq 'iso-10646-comment attributes)
1126                (setq value (get-char-attribute char 'iso-10646-comment)))
1127       (insert (format "(iso-10646-comment\t. %S)%s"
1128                       value
1129                       line-breaking))
1130       (setq attributes (delq 'iso-10646-comment attributes))
1131       )
1132     (when (and (memq 'morohashi-daikanwa attributes)
1133                (setq value (get-char-attribute char 'morohashi-daikanwa)))
1134       (insert (format "(morohashi-daikanwa\t%s)%s"
1135                       (mapconcat (function prin1-to-string) value " ")
1136                       line-breaking))
1137       (setq attributes (delq 'morohashi-daikanwa attributes))
1138       )
1139     (setq radical nil
1140           strokes nil)
1141     (when (and (memq 'ideographic-radical attributes)
1142                (setq value (get-char-attribute char 'ideographic-radical)))
1143       (setq radical value)
1144       (insert (format "(ideographic-radical . %S)\t; %c%s"
1145                       radical
1146                       (ideographic-radical radical)
1147                       line-breaking))
1148       (setq attributes (delq 'ideographic-radical attributes))
1149       )
1150     (when (and (memq 'shuowen-radical attributes)
1151                (setq value (get-char-attribute char 'shuowen-radical)))
1152       (insert (format "(shuowen-radical\t. %S)\t; %c%s"
1153                       value
1154                       (shuowen-radical value)
1155                       line-breaking))
1156       (setq attributes (delq 'shuowen-radical attributes))
1157       )
1158     (let (key)
1159       (dolist (domain
1160                (append
1161                 char-db-feature-domains
1162                 (let (dest domain)
1163                   (dolist (feature (char-attribute-list))
1164                     (setq feature (symbol-name feature))
1165                     (when (string-match
1166                            "\\(radical\\|strokes\\)@\\([^@*]+\\)\\(\\*\\|$\\)"
1167                            feature)
1168                       (setq domain (intern (match-string 2 feature)))
1169                      (unless (memq domain dest)
1170                        (setq dest (cons domain dest)))))
1171                   (sort dest #'string<))))
1172         (setq key (intern (format "%s@%s" 'ideographic-radical domain)))
1173         (when (and (memq key attributes)
1174                    (setq value (get-char-attribute char key)))
1175           (setq radical value)
1176           (insert (format "(%s . %S)\t; %c%s"
1177                           key
1178                           radical
1179                           (ideographic-radical radical)
1180                           line-breaking))
1181           (setq attributes (delq key attributes))
1182           )
1183         (setq key (intern (format "%s@%s" 'ideographic-strokes domain)))
1184         (when (and (memq key attributes)
1185                    (setq value (get-char-attribute char key)))
1186           (setq strokes value)
1187           (insert (format "(%s . %S)%s"
1188                           key
1189                           strokes
1190                           line-breaking))
1191           (setq attributes (delq key attributes))
1192           )
1193         (setq key (intern (format "%s@%s" 'total-strokes domain)))
1194         (when (and (memq key attributes)
1195                    (setq value (get-char-attribute char key)))
1196           (insert (format "(%s       . %S)%s"
1197                           key
1198                           value
1199                           line-breaking))
1200           (setq attributes (delq key attributes))
1201           )
1202         (dolist (feature '(ideographic-radical
1203                            ideographic-strokes
1204                            total-strokes))
1205           (setq key (intern (format "%s@%s*sources" feature domain)))
1206           (when (and (memq key attributes)
1207                      (setq value (get-char-attribute char key)))
1208             (insert (format "(%s%s" key line-breaking))
1209             (dolist (cell value)
1210               (insert (format " %s" cell)))
1211             (insert ")")
1212             (insert line-breaking)
1213             (setq attributes (delq key attributes))
1214             ))
1215         ))
1216     (when (and (memq 'ideographic-strokes attributes)
1217                (setq value (get-char-attribute char 'ideographic-strokes)))
1218       (setq strokes value)
1219       (insert (format "(ideographic-strokes . %S)%s"
1220                       strokes
1221                       line-breaking))
1222       (setq attributes (delq 'ideographic-strokes attributes))
1223       )
1224     (when (and (memq 'kangxi-radical attributes)
1225                (setq value (get-char-attribute char 'kangxi-radical)))
1226       (unless (eq value radical)
1227         (insert (format "(kangxi-radical\t . %S)\t; %c%s"
1228                         value
1229                         (ideographic-radical value)
1230                         line-breaking))
1231         (or radical
1232             (setq radical value)))
1233       (setq attributes (delq 'kangxi-radical attributes))
1234       )
1235     (when (and (memq 'kangxi-strokes attributes)
1236                (setq value (get-char-attribute char 'kangxi-strokes)))
1237       (unless (eq value strokes)
1238         (insert (format "(kangxi-strokes\t . %S)%s"
1239                         value
1240                         line-breaking))
1241         (or strokes
1242             (setq strokes value)))
1243       (setq attributes (delq 'kangxi-strokes attributes))
1244       )
1245     ;; (when (and (memq 'japanese-radical attributes)
1246     ;;            (setq value (get-char-attribute char 'japanese-radical)))
1247     ;;   (unless (eq value radical)
1248     ;;     (insert (format "(japanese-radical\t . %S)\t; %c%s"
1249     ;;                     value
1250     ;;                     (ideographic-radical value)
1251     ;;                     line-breaking))
1252     ;;     (or radical
1253     ;;         (setq radical value)))
1254     ;;   (setq attributes (delq 'japanese-radical attributes))
1255     ;;   )
1256     (when (and (memq 'japanese-strokes attributes)
1257                (setq value (get-char-attribute char 'japanese-strokes)))
1258       (unless (eq value strokes)
1259         (insert (format "(japanese-strokes\t . %S)%s"
1260                         value
1261                         line-breaking))
1262         (or strokes
1263             (setq strokes value)))
1264       (setq attributes (delq 'japanese-strokes attributes))
1265       )
1266     (when (and (memq 'cns-radical attributes)
1267                (setq value (get-char-attribute char 'cns-radical)))
1268       (insert (format "(cns-radical\t . %S)\t; %c%s"
1269                       value
1270                       (ideographic-radical value)
1271                       line-breaking))
1272       (setq attributes (delq 'cns-radical attributes))
1273       )
1274     (when (and (memq 'cns-strokes attributes)
1275                (setq value (get-char-attribute char 'cns-strokes)))
1276       (unless (eq value strokes)
1277         (insert (format "(cns-strokes\t . %S)%s"
1278                         value
1279                         line-breaking))
1280         (or strokes
1281             (setq strokes value)))
1282       (setq attributes (delq 'cns-strokes attributes))
1283       )
1284     ;; (when (and (memq 'ideographic- attributes)
1285     ;;            (setq value (get-char-attribute char 'ideographic-)))
1286     ;;   (insert "(ideographic-       ")
1287     ;;   (setq lbs (concat "\n" (make-string (current-column) ?\ ))
1288     ;;         separator nil)
1289     ;;   (while (consp value)
1290     ;;     (setq cell (car value))
1291     ;;     (if (integerp cell)
1292     ;;         (setq cell (decode-char '=ucs cell)))
1293     ;;     (cond ((characterp cell)
1294     ;;            (if separator
1295     ;;                (insert lbs))
1296     ;;            (if readable
1297     ;;                (insert (format "%S" cell))
1298     ;;              (char-db-insert-char-spec cell readable))
1299     ;;            (setq separator lbs))
1300     ;;           ((consp cell)
1301     ;;            (if separator
1302     ;;                (insert lbs))
1303     ;;            (if (consp (car cell))
1304     ;;                (char-db-insert-char-spec cell readable)
1305     ;;              (char-db-insert-char-reference cell readable))
1306     ;;            (setq separator lbs))
1307     ;;           (t
1308     ;;            (if separator
1309     ;;                (insert separator))
1310     ;;            (insert (prin1-to-string cell))
1311     ;;            (setq separator " ")))
1312     ;;     (setq value (cdr value)))
1313     ;;   (insert ")")
1314     ;;   (insert line-breaking)
1315     ;;   (setq attributes (delq 'ideographic- attributes)))
1316     (when (and (memq 'total-strokes attributes)
1317                (setq value (get-char-attribute char 'total-strokes)))
1318       (insert (format "(total-strokes       . %S)%s"
1319                       value
1320                       line-breaking))
1321       (setq attributes (delq 'total-strokes attributes))
1322       )
1323     ;; (when (and (memq '->ideograph attributes)
1324     ;;            (setq value (get-char-attribute char '->ideograph)))
1325     ;;   (insert (format "(->ideograph\t%s)%s"
1326     ;;                   (mapconcat (lambda (code)
1327     ;;                                (cond ((symbolp code)
1328     ;;                                       (symbol-name code))
1329     ;;                                      ((integerp code)
1330     ;;                                       (format "#x%04X" code))
1331     ;;                                      (t
1332     ;;                                       (format "%s %S"
1333     ;;                                               line-breaking code))))
1334     ;;                              value " ")
1335     ;;                   line-breaking))
1336     ;;   (setq attributes (delq '->ideograph attributes))
1337     ;;   )
1338     ;; (when (and (memq '->decomposition attributes)
1339     ;;            (setq value (get-char-attribute char '->decomposition)))
1340     ;;   (insert (format "(->decomposition\t%s)%s"
1341     ;;                   (mapconcat (lambda (code)
1342     ;;                                (cond ((symbolp code)
1343     ;;                                       (symbol-name code))
1344     ;;                                      ((characterp code)
1345     ;;                                       (if readable
1346     ;;                                           (format "%S" code)
1347     ;;                                         (format "#x%04X"
1348     ;;                                                 (char-int code))
1349     ;;                                         ))
1350     ;;                                      ((integerp code)
1351     ;;                                       (format "#x%04X" code))
1352     ;;                                      (t
1353     ;;                                       (format "%s%S" line-breaking code))))
1354     ;;                              value " ")
1355     ;;                   line-breaking))
1356     ;;   (setq attributes (delq '->decomposition attributes))
1357     ;;   )
1358     (if (equal (get-char-attribute char '->titlecase)
1359                (get-char-attribute char '->uppercase))
1360         (setq attributes (delq '->titlecase attributes)))
1361     (when (and (memq '->mojikyo attributes)
1362                (setq value (get-char-attribute char '->mojikyo)))
1363       (insert (format "(->mojikyo\t\t. %06d)\t; %c%s"
1364                       value (decode-char 'mojikyo value)
1365                       line-breaking))
1366       (setq attributes (delq '->mojikyo attributes))
1367       )
1368     (unless readable
1369       (dolist (ignored '(composition
1370                          ->denotational <-subsumptive ->ucs-unified
1371                          ;; ->ideographic-component-forms
1372                          ))
1373         (setq attributes (delq ignored attributes))))
1374     (while attributes
1375       (setq name (car attributes))
1376       (unless (eq (setq value (get-char-attribute char name 'value-is-empty))
1377                   'value-is-empty)
1378         (cond ((setq ret (find-charset name))
1379                (setq name (charset-name ret))
1380                (when (not (memq name dest-ccss))
1381                  (setq dest-ccss (cons name dest-ccss))
1382                  (char-db-insert-ccs-feature name value line-breaking))
1383                )
1384               ((string-match "^=>ucs@" (symbol-name name))
1385                (insert (format "(%-18s . #x%04X)\t; %c%s"
1386                                name value (decode-char '=ucs value)
1387                                line-breaking))
1388                )
1389               ((eq name 'jisx0208-1978/4X)
1390                (insert (format "(%-18s . #x%04X)%s"
1391                                name value
1392                                line-breaking))
1393                )
1394               ((and
1395                 (not readable)
1396                 (not (eq name '->subsumptive))
1397                 (not (eq name '->uppercase))
1398                 (not (eq name '->lowercase))
1399                 (not (eq name '->titlecase))
1400                 (not (eq name '->canonical))
1401                 (not (eq name '->Bopomofo))
1402                 (not (eq name '->mistakable))
1403                 (not (eq name '->ideographic-variants))
1404                 (null (get-char-attribute
1405                        char (intern (format "%s*sources" name))))
1406                 (not (string-match "\\*sources$" (symbol-name name)))
1407                 (null (get-char-attribute
1408                        char (intern (format "%s*note" name))))
1409                 (not (string-match "\\*note$" (symbol-name name)))
1410                 (or (eq name '<-identical)
1411                     (eq name '<-uppercase)
1412                     (eq name '<-lowercase)
1413                     (eq name '<-titlecase)
1414                     (eq name '<-canonical)
1415                     (eq name '<-ideographic-variants)
1416                     ;; (eq name '<-synonyms)
1417                     (string-match "^<-synonyms" (symbol-name name))
1418                     (eq name '<-mistakable)
1419                     (when (string-match "^->" (symbol-name name))
1420                       (cond
1421                        ((string-match "^->fullwidth" (symbol-name name))
1422                         (not (and (consp value)
1423                                   (characterp (car value))
1424                                   (encode-char
1425                                    (car value) '=ucs 'defined-only)))
1426                         )
1427                        (t)))
1428                     ))
1429                )
1430               ((or (eq name 'ideographic-structure)
1431                    (eq name 'ideographic-structure@apparent)
1432                    (eq name 'ideographic-structure@apparent/leftmost)
1433                    (eq name 'ideographic-combination)
1434                    ;; (eq name 'ideographic-)
1435                    (eq name '=decomposition)
1436                    (char-feature-base-name= '=decomposition name)
1437                    (char-feature-base-name= '=>decomposition name)
1438                    ;; (string-match "^=>*decomposition\\(@[^*]+\\)?$"
1439                    ;;               (symbol-name name))
1440                    (string-match "^\\(->\\|<-\\)[^*]*$" (symbol-name name))
1441                    (string-match "^\\(->\\|<-\\)[^*]*\\*sources$"
1442                                  (symbol-name name))
1443                    )
1444                (char-db-insert-relation-feature char name value
1445                                                 line-breaking
1446                                                 ccss readable))
1447               ;; ((memq name '(ideograph=
1448               ;;               original-ideograph-of
1449               ;;               ancient-ideograph-of
1450               ;;               vulgar-ideograph-of
1451               ;;               wrong-ideograph-of
1452               ;;               ;; simplified-ideograph-of
1453               ;;               ideographic-variants
1454               ;;               ;; ideographic-different-form-of
1455               ;;               ))
1456               ;;  (insert (format "(%-18s%s " name line-breaking))
1457               ;;  (setq lbs (concat "\n" (make-string (current-column) ?\ ))
1458               ;;        separator nil)
1459               ;;  (while (consp value)
1460               ;;    (setq cell (car value))
1461               ;;    (if (and (consp cell)
1462               ;;             (consp (car cell)))
1463               ;;        (progn
1464               ;;          (if separator
1465               ;;              (insert lbs))
1466               ;;          (char-db-insert-alist cell readable)
1467               ;;          (setq separator lbs))
1468               ;;      (if separator
1469               ;;          (insert separator))
1470               ;;      (insert (prin1-to-string cell))
1471               ;;      (setq separator " "))
1472               ;;    (setq value (cdr value)))
1473               ;;  (insert ")")
1474               ;;  (insert line-breaking))
1475               ((consp value)
1476                (insert (format "(%-18s " name))
1477                (setq lbs (concat "\n" (make-string (current-column) ?\ ))
1478                      separator nil)
1479                (while (consp value)
1480                  (setq cell (car value))
1481                  (if (and (consp cell)
1482                           (consp (car cell))
1483                           (setq ret (condition-case nil
1484                                         (find-char cell)
1485                                       (error nil))))
1486                      (progn
1487                        (setq rest cell
1488                              al nil
1489                              cal nil)
1490                        (while rest
1491                          (setq key (car (car rest)))
1492                          (if (find-charset key)
1493                              (setq cal (cons key cal))
1494                            (setq al (cons key al)))
1495                          (setq rest (cdr rest)))
1496                        (if separator
1497                            (insert lbs))
1498                        (insert-char-attributes ret
1499                                                readable
1500                                                al ; cal
1501                                                nil 'for-sub-node)
1502                        (setq separator lbs))
1503                    (setq ret (prin1-to-string cell))
1504                    (if separator
1505                        (if (< (+ (current-column)
1506                                  (length ret)
1507                                  (length separator))
1508                               76)
1509                            (insert separator)
1510                          (insert lbs)))
1511                    (insert ret)
1512                    (setq separator " "))
1513                  (setq value (cdr value)))
1514                (insert ")")
1515                (insert line-breaking))
1516               (t
1517                (insert (format "(%-18s" name))
1518                (setq ret (prin1-to-string value))
1519                (unless (< (+ (current-column)
1520                              (length ret)
1521                              3)
1522                           76)
1523                  (insert line-breaking))
1524                (insert " . " ret ")" line-breaking)
1525                ;; (insert (format "(%-18s . %S)%s"
1526                ;;                 name value
1527                ;;                 line-breaking))
1528                )
1529               ))
1530       (setq attributes (cdr attributes)))
1531     (insert ")")))
1532
1533 (defun insert-char-data (char &optional readable
1534                               attributes)
1535   (save-restriction
1536     (narrow-to-region (point)(point))
1537     (insert "(define-char
1538   '")
1539     (insert-char-attributes char readable attributes)
1540     (insert ")\n")
1541     (goto-char (point-min))
1542     (while (re-search-forward "[ \t]+$" nil t)
1543       (replace-match ""))
1544     ;; from tabify.
1545     (goto-char (point-min))
1546     (while (re-search-forward "[ \t][ \t][ \t]*" nil t)
1547       (let ((column (current-column))
1548             (indent-tabs-mode t))
1549         (delete-region (match-beginning 0) (point))
1550         (indent-to column)))
1551     (goto-char (point-max))
1552     ;; (tabify (point-min)(point-max))
1553     ))
1554
1555 (defun insert-char-data-with-variant (char &optional printable
1556                                            no-ucs-unified
1557                                            script excluded-script)
1558   (insert-char-data char printable)
1559   (let ((variants (char-variants char))
1560         rest
1561         variant vs ret)
1562     (setq variants (sort variants #'<))
1563     (setq rest variants)
1564     (setq variants (cons char variants))
1565     (while rest
1566       (setq variant (car rest))
1567       (unless (get-char-attribute variant '<-subsumptive)
1568         (if (and (or (null script)
1569                      (null (setq vs (get-char-attribute variant 'script)))
1570                      (memq script vs))
1571                  (or (null excluded-script)
1572                      (null (setq vs (get-char-attribute variant 'script)))
1573                      (not (memq excluded-script vs))))
1574             (unless (and no-ucs-unified (get-char-attribute variant '=ucs))
1575               (insert-char-data variant printable)
1576               (if (setq ret (char-variants variant))
1577                   (while ret
1578                     (or (memq (car ret) variants)
1579                         ;; (get-char-attribute (car ret) '<-subsumptive)
1580                         (setq rest (nconc rest (list (car ret)))))
1581                     (setq ret (cdr ret)))))))
1582       (setq rest (cdr rest)))))
1583
1584 (defun insert-char-range-data (min max &optional script excluded-script)
1585   (let ((code min)
1586         char)
1587     (while (<= code max)
1588       (setq char (decode-char '=ucs code))
1589       (if (encode-char char '=ucs 'defined-only)
1590           (insert-char-data-with-variant char nil 'no-ucs-unified
1591                                          script excluded-script))
1592       (setq code (1+ code)))))
1593
1594 (defun write-char-range-data-to-file (min max file
1595                                           &optional script excluded-script)
1596   (let ((coding-system-for-write char-db-file-coding-system))
1597     (with-temp-buffer
1598       (insert (format ";; -*- coding: %s -*-\n"
1599                       char-db-file-coding-system))
1600       (insert-char-range-data min max script excluded-script)
1601       (write-region (point-min)(point-max) file))))
1602
1603 (defvar what-character-original-window-configuration)
1604
1605 ;;;###autoload
1606 (defun what-char-definition (char)
1607   (interactive (list (char-after)))
1608   (let ((buf (get-buffer-create "*Character Description*"))
1609         (the-buf (current-buffer))
1610         (win-conf (current-window-configuration)))
1611     (pop-to-buffer buf)
1612     (make-local-variable 'what-character-original-window-configuration)
1613     (setq what-character-original-window-configuration win-conf)
1614     (setq buffer-read-only nil)
1615     (erase-buffer)
1616     (condition-case err
1617         (progn
1618           (insert-char-data-with-variant char 'printable)
1619           (unless (char-attribute-alist char)
1620             (insert (format ";; = %c\n"
1621                             (let* ((rest (split-char char))
1622                                    (ccs (pop rest))
1623                                    (code (pop rest)))
1624                               (while rest
1625                                 (setq code (logior (lsh code 8)
1626                                                    (pop rest))))
1627                               (decode-char ccs code)))))
1628           ;; (char-db-update-comment)
1629           (set-buffer-modified-p nil)
1630           (view-mode the-buf (lambda (buf)
1631                                (set-window-configuration
1632                                 what-character-original-window-configuration)
1633                                ))
1634           (goto-char (point-min)))
1635       (error (progn
1636                (set-window-configuration
1637                 what-character-original-window-configuration)
1638                (signal (car err) (cdr err)))))))
1639
1640
1641 ;;; @ end
1642 ;;;
1643
1644 (provide 'char-db-util)
1645
1646 ;;; char-db-util.el ends here