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

   1 ;;; char-db-util.el --- Character Database utility -*- coding: utf-8-er; -*-
   2
   3 ;; Copyright (C) 1998, 1999, 2000, 2001, 2002, 2003, 2004, 2005, 2006, 2007,
   4 ;;   2008, 2009, 2010, 2011, 2012, 2013, 2014, 2015, 2016, 2017, 2018, 2019,
   5 ;;   2020, 2021 MORIOKA Tomohiko.
   6
   7 ;; Author: MORIOKA Tomohiko <tomo@kanji.zinbun.kyoto-u.ac.jp>
   8 ;; Keywords: CHISE, Character Database, ISO/IEC 10646, UCS, Unicode, MULE.
   9
  10 ;; This file is part of XEmacs CHISE.
  11
  12 ;; XEmacs CHISE is free software; you can redistribute it and/or
  13 ;; modify it under the terms of the GNU General Public License as
  14 ;; published by the Free Software Foundation; either version 2, or (at
  15 ;; your option) any later version.
  16
  17 ;; XEmacs CHISE is distributed in the hope that it will be useful,
  18 ;; but WITHOUT ANY WARRANTY; without even the implied warranty of
  19 ;; MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the GNU
  20 ;; General Public License for more details.
  21
  22 ;; You should have received a copy of the GNU General Public License
  23 ;; along with XEmacs CHISE; see the file COPYING.  If not, write to
  24 ;; the Free Software Foundation, Inc., 59 Temple Place - Suite 330,
  25 ;; Boston, MA 02111-1307, USA.
  26
  27 ;;; Code:
  28
  29 (require 'chise-subr)
  30 (require 'ideograph-subr)
  31
  32 (defconst unidata-normative-category-alist
  33   '(("Lu" letter        uppercase)
  34     ("Ll" letter        lowercase)
  35     ("Lt" letter        titlecase)
  36     ("Mn" mark          non-spacing)
  37     ("Mc" mark          spacing-combining)
  38     ("Me" mark          enclosing)
  39     ("Nd" number        decimal-digit)
  40     ("Nl" number        letter)
  41     ("No" number        other)
  42     ("Zs" separator     space)
  43     ("Zl" separator     line)
  44     ("Zp" separator     paragraph)
  45     ("Cc" other         control)
  46     ("Cf" other         format)
  47     ("Cs" other         surrogate)
  48     ("Co" other         private-use)
  49     ("Cn" other         not-assigned)))
  50
  51 (defconst unidata-informative-category-alist
  52   '(("Lm" letter        modifier)
  53     ("Lo" letter        other)
  54     ("Pc" punctuation   connector)
  55     ("Pd" punctuation   dash)
  56     ("Ps" punctuation   open)
  57     ("Pe" punctuation   close)
  58     ("Pi" punctuation   initial-quote)
  59     ("Pf" punctuation   final-quote)
  60     ("Po" punctuation   other)
  61     ("Sm" symbol        math)
  62     ("Sc" symbol        currency)
  63     ("Sk" symbol        modifier)
  64     ("So" symbol        other)
  65     ))
  66
  67 (defconst shuowen-radicals
  68   [?一 ?上 ?示 ?三 ?王 ?玉 ?玨 ?气 ?士 ?丨 ; 010
  69    ?屮 ?艸 ?蓐 ?茻 ?小 ?八 ?釆 ?半 ?牛 ?犛 ; 020
  70    ?告 ?口 ?凵 ?吅 ?哭 ?走 ?止 ?癶 ?步 ?此 ; 030
  71    ?正 ?是 ?辵 ?彳 ?廴 ?㢟 ?行 ?齒 ?牙 ?足 ; 040
  72    ?疋 ?品 ?龠 ?冊 ?㗊 ?舌 ?干 ?𧮫 ?只 ?㕯 ; 050
  73    ?句 ?丩 ?古 ?十 ?卅 ?言 ?誩 ?音 ?䇂 ?丵 ; 060
  74    ?菐 ?𠬞 ?𠬜 ?共 ?異 ?舁 ?𦥑 ?䢅 ?爨 ?革 ; 070
  75    ?鬲 ?䰜 ?爪 ?𠃨 ?鬥 ?又 ?𠂇 ?史 ?支 ?𦘒 ; 080
  76    ?聿 ?畫 ?隶 ?臤 ?臣 ?殳 ?殺 ?𠘧 ?寸 ?皮 ; 090
  77    ?㼱 ?攴 ?敎 ?卜 ?用 ?爻 ?㸚 ?𥄎 ?目 ?䀠 ; 100
  78    ?眉 ?盾 ?自 ?白 ?鼻 ?皕 ?習 ?羽 ?隹 ?奞 ; 110
  79    ?雈 ?𦫳 ?𥄕 ?羊 ?羴 ?瞿 ?雔 ?雥 ?鳥 ?烏 ; 120
  80    ?𠦒 ?冓 ?幺 ?𢆶 ?叀 ?玄 ?予 ?放 ?𠬪 ?𣦼 ; 130
  81    ?歺 ?死 ?冎 ?骨 ?肉 ?筋 ?刀 ?刃 ?㓞 ?丯 ; 140
  82    ?耒 ?角 ?竹 ?箕 ?丌 ?左 ?工 ?㠭 ?巫 ?甘 ; 150
  83    ?曰 ?乃 ?丂 ?可 ?兮 ?号 ?亏 ?旨 ?喜 ?壴 ; 160
  84 ;  ?旨 ?曰 ?乃 ?丂 ?可 ?兮 ?号 ?亏 ?喜 ?壴 ; 160
  85    ?鼓 ?豈 ?豆 ?豊 ?豐 ?䖒 ?虍 ?虎 ?虤 ?皿 ; 170
  86    ?𠙴 ?去 ?血 ?丶 ?丹 ?青 ?井 ?皀 ?鬯 ?食 ; 180
  87    ?亼 ?會 ?倉 ?入 ?缶 ?矢 ?高 ?冂 ?𩫏 ?京 ; 190
  88    ?亯 ?𣆪 ?畗 ?㐭 ?嗇 ?來 ?麥 ?夊 ?舛 ?䑞 ; 200
  89    ?韋 ?弟 ?夂 ?久 ?桀 ?木 ?東 ?林 ?才 ?叒 ; 210
  90    ?之 ?帀 ?出 ?𣎵 ?生 ?乇 ?𠂹 ?𠌶 ?華 ?𥝌 ; 220
  91    ?稽 ?巢 ?桼 ?束 ?㯻 ?囗 ?員 ?貝 ?邑 ?𨛜 ; 230
  92    ?日 ?旦 ?倝 ?㫃 ?冥 ?晶 ?月 ?有 ?明 ?囧 ; 240
  93    ?夕 ?多 ?毌 ?𢎘 ?𣐺 ?卣 ?齊 ?朿 ?片 ?鼎 ; 250
  94    ?克 ?彔 ?禾 ?秝 ?黍 ?香 ?米 ?毇 ?臼 ?凶 ; 260
  95    ?𣎳 ?𣏟 ?麻 ?尗 ?耑 ?韭 ?瓜 ?瓠 ?宀 ?宮 ; 270
  96    ?呂 ?穴 ?㝱 ?疒 ?冖 ?𠔼 ?冃 ?㒳 ?网 ?襾 ; 280
  97    ?巾 ?巿 ?帛 ?白 ?㡀 ?黹 ?人 ?𠤎 ?匕 ?从 ; 290
  98    ?比 ?北 ?丘 ?㐺 ?𡈼 ?重 ?臥 ?身 ?㐆 ?衣 ; 300
  99    ?裘 ?老 ?毛 ?毳 ?尸 ?尺 ?尾 ?履 ?舟 ?方 ; 310
 100    ?儿 ?兄 ?兂 ?皃 ?𠑹 ?先 ?秃 ?見 ?覞 ?欠 ; 320
 101    ?㱃 ?㳄 ?旡 ?頁 ?𦣻 ?面 ?丏 ?首 ?𥄉 ?須 ; 330
 102    ?彡 ?彣 ?文 ?髟 ?后 ?司 ?卮 ?卩 ?印 ?色 ; 340
 103    ?𠨍 ?辟 ?勹 ?包 ?茍 ?鬼 ?甶 ?厶 ?嵬 ?山 ; 350
 104    ?屾 ?屵 ?广 ?厂 ?丸 ?危 ?石 ?長 ?勿 ?冄 ; 360
 105    ?而 ?豕 ?㣇 ?彑 ?豚 ?豸 ?𤉡 ?易 ?象 ?馬 ; 370
 106    ?𢊁 ?鹿 ?麤 ?㲋 ?兔 ?萈 ?犬 ?㹜 ?鼠 ?能 ; 380
 107    ?熊 ?火 ?炎 ?黑 ?囪 ?焱 ?炙 ?赤 ?大 ?亦 ; 390
 108    ?夨 ?夭 ?交 ?尣 ?壺 ?壹 ?㚔 ?奢 ?亢 ?夲 ; 400
 109    ?夰 ?亣 ?夫 ?立 ?竝 ?囟 ?思 ?心 ?惢 ?水 ; 410
 110    ?沝 ?瀕 ?𡿨 ?巜 ?川 ?泉 ?灥 ?永 ?𠂢 ?谷 ; 420
 111    ?仌 ?雨 ?雲 ?魚 ?𩺰 ?燕 ?龍 ?飛 ?非 ?卂 ; 430
 112    ?𠃉 ?不 ?至 ?西 ?鹵 ?鹽 ?戶 ?門 ?耳 ?𦣝 ; 440
 113    ?手 ?𠦬 ?女 ?毋 ?民 ?丿 ?𠂆 ?乁 ?氏 ?氐 ; 450
 114    ?戈 ?戉 ?我 ?亅 ?琴 ?𠃊 ?亡 ?匸 ?匚 ?曲 ; 460
 115    ?甾 ?瓦 ?弓 ?弜 ?弦 ?系 ?糸 ?素 ?絲 ?率 ; 470
 116    ?虫 ?䖵 ?蟲 ?風 ?它 ?龜 ?黽 ?卵 ?二 ?土 ; 480
 117    ?垚 ?堇 ?里 ?田 ?畕 ?黃 ?男 ?力 ?劦 ?金 ; 490
 118    ?幵 ?勺 ?几 ?且 ?斤 ?斗 ?矛 ?車 ?𠂤 ?𨸏 ; 500
 119    ?𨺅 ?厽 ?四 ?宁 ?叕 ?亞 ?五 ?六 ?七 ?九 ; 510
 120    ?禸 ?嘼 ?甲 ?乙 ?丙 ?丁 ?戊 ?己 ?巴 ?庚 ; 520
 121    ?辛 ?辡 ?壬 ?癸 ?子 ?了 ?孨 ?𠫓 ?丑 ?寅 ; 530
 122    ?卯 ?辰 ?巳 ?午 ?未 ?申 ?酉 ?酋 ?戌 ?亥 ; 540
 123    ])
 124
 125 (defun shuowen-radical (number)
 126   (aref shuowen-radicals (1- number)))
 127
 128 (defvar char-db-file-coding-system 'utf-8-mcs-er)
 129
 130 (defvar char-db-ignored-attributes '(ideographic-products))
 131
 132 (defvar char-db-coded-charset-priority-list
 133   '(ascii
 134     control-1
 135     latin-iso8859-1
 136     latin-iso8859-2
 137     latin-iso8859-3
 138     latin-iso8859-4
 139     latin-iso8859-9
 140     latin-jisx0201
 141     cyrillic-iso8859-5
 142     greek-iso8859-7
 143     thai-tis620
 144     ;; =mj
 145     ;; =adobe-japan1-0
 146     ;; =adobe-japan1-1
 147     ;; =adobe-japan1-2
 148     ;; =adobe-japan1-3
 149     ;; =adobe-japan1-4
 150     ;; =adobe-japan1-5
 151     ;; =adobe-japan1-6
 152     =jis-x0208
 153     =jis-x0208@1978
 154     =jis-x0208@1983
 155     japanese-jisx0212
 156     chinese-gb2312
 157     =jis-x0208@1990
 158     chinese-cns11643-1
 159     chinese-cns11643-2
 160     chinese-cns11643-3
 161     chinese-cns11643-4
 162     chinese-cns11643-5
 163     chinese-cns11643-6
 164     chinese-cns11643-7
 165     =jis-x0213-1
 166     =jis-x0213-1@2000
 167     =jis-x0213-1@2004
 168     =jis-x0213-2
 169     korean-ksc5601
 170     chinese-isoir165
 171     katakana-jisx0201
 172     hebrew-iso8859-8
 173     chinese-gb12345
 174     latin-viscii
 175     ethiopic-ucs
 176     =big5-cdp
 177     =hanyo-denshi/ja
 178     =hanyo-denshi/jb
 179     =hanyo-denshi/jc
 180     =hanyo-denshi/jd
 181     =hanyo-denshi/ft
 182     =hanyo-denshi/ia
 183     =hanyo-denshi/ib
 184     =hanyo-denshi/hg
 185     =hanyo-denshi/jt
 186     =hanyo-denshi/ks
 187     =hanyo-denshi/tk
 188     ;; ==mj
 189     ;; ==adobe-japan1-0
 190     ;; ==adobe-japan1-1
 191     ;; ==adobe-japan1-2
 192     ;; ==adobe-japan1-3
 193     ;; ==adobe-japan1-4
 194     ;; ==adobe-japan1-5
 195     ;; ==adobe-japan1-6
 196     ==jis-x0208
 197     ==jis-x0213-1
 198     ==jis-x0213-2
 199     ==jis-x0212
 200     ==hanyo-denshi/ja
 201     ==hanyo-denshi/jb
 202     ==hanyo-denshi/jc
 203     ==hanyo-denshi/ft
 204     ==hanyo-denshi/ia
 205     ==hanyo-denshi/ib
 206     ==hanyo-denshi/hg
 207     ==hanyo-denshi/jt
 208     ==hanyo-denshi/ks
 209     ==hanyo-denshi/tk
 210     =gt
 211     =gt-k
 212     =daikanwa
 213     =daijiten
 214     =daikanwa@rev2
 215     =daikanwa@rev1
 216     =daikanwa/+p
 217     =daikanwa/ho
 218     ==daikanwa
 219     ==daijiten
 220     =cbeta
 221     ideograph-hanziku-1
 222     ideograph-hanziku-2
 223     ideograph-hanziku-3
 224     ideograph-hanziku-4
 225     ideograph-hanziku-5
 226     ideograph-hanziku-6
 227     ideograph-hanziku-7
 228     ideograph-hanziku-8
 229     ideograph-hanziku-9
 230     ideograph-hanziku-10
 231     ideograph-hanziku-11
 232     ideograph-hanziku-12
 233     ;; =>>>adobe-japan1-0
 234     ;; =>>>adobe-japan1-1
 235     ;; =>>>adobe-japan1-2
 236     ;; =>>>adobe-japan1-3
 237     ;; =>>>adobe-japan1-4
 238     ;; =>>>adobe-japan1-5
 239     ;; =>>>adobe-japan1-6
 240     ;; =>>>jis-x0208
 241     ;; =>>>jis-x0213-1
 242     ;; =>>>jis-x0213-2
 243     ;; =>>>hanyo-denshi/ja
 244     ;; =>>>hanyo-denshi/jb
 245     ;; =>>>hanyo-denshi/jc
 246     ;; =>>>hanyo-denshi/ft
 247     ;; =>>>hanyo-denshi/ib
 248     ;; =>>>hanyo-denshi/hg
 249     ;; =>>>hanyo-denshi/jt
 250     ;; =>>>hanyo-denshi/ks
 251     ;; =>>>gt
 252     =>>adobe-japan1-0
 253     =>>adobe-japan1-1
 254     =>>adobe-japan1-2
 255     =>>adobe-japan1-3
 256     =>>adobe-japan1-4
 257     =>>adobe-japan1-5
 258     =>>adobe-japan1-6
 259     =>>jis-x0208
 260     =>>jis-x0213-1
 261     =>>jis-x0213-1@2000
 262     =>>jis-x0213-1@2004
 263     =>>jis-x0213-2
 264     =>>jis-x0208@1978
 265     =>>hanyo-denshi/ft
 266     =>>hanyo-denshi/jt
 267     =>>hanyo-denshi/ks
 268     =>>gt
 269     =>>daikanwa
 270     =>>cbeta
 271     =+>jis-x0208
 272     =+>jis-x0213-1
 273     =+>jis-x0213-2
 274     =+>adobe-japan1-0
 275     =+>adobe-japan1-1
 276     =+>adobe-japan1-2
 277     =+>adobe-japan1-3
 278     =+>adobe-japan1-4
 279     =+>adobe-japan1-5
 280     =+>adobe-japan1-6
 281     =+>jis-x0208@1978
 282     =>jis-x0208
 283     =>jis-x0208@1997
 284     =>jis-x0213-1
 285     =>jis-x0213-1@2000
 286     =>jis-x0213-1@2004
 287     =>jis-x0213-2
 288     ==>ucs@bucs
 289     ==>daijiten
 290     ;; =>ucs@hanyo-denshi
 291     =>ucs@iso
 292     =>ucs@unicode
 293     =>ucs@jis
 294     =>ucs@cns
 295     =>ucs@ks
 296     =+>ucs@iso
 297     =+>ucs@unicode
 298     =+>ucs@jis
 299     =+>ucs@jis/1990
 300     =+>ucs@cns
 301     =+>ucs@ks
 302     =>>ucs@iso
 303     =>>ucs@unicode
 304     =>>ucs@jis
 305     =>>ucs@cns
 306     =>>>ucs@iso
 307     =>>>ucs@unicode
 308     ==ucs@iso
 309     ==ucs@unicode
 310     ;; ==ucs@cns
 311     ==gb2312
 312     ==ks-x1001
 313     ==cns11643-1
 314     ==cns11643-2
 315     ==cns11643-3
 316     ==cns11643-4
 317     ==cns11643-5
 318     ==cns11643-6
 319     ==cns11643-7
 320     ==gt
 321     ==jis-x0208@1990
 322     ;; ==jis-x0208@1983
 323     ==jis-x0208@1978
 324     ==gt-k
 325     =ucs@iso
 326     =ucs@unicode
 327     =ucs@cns
 328     ==big5-cdp
 329     ==cbeta
 330     =>>big5-cdp
 331     =>>gt-k
 332     =+>gt
 333     =+>big5-cdp
 334     =>big5-cdp
 335     =>big5-cdp@iwds-1
 336     =>mj
 337     =>iwds-1
 338     =big5
 339     =big5-eten
 340     =zinbun-oracle
 341     =>zinbun-oracle
 342     =ruimoku-v6
 343     =>>ruimoku-v6
 344     ==ruimoku-v6
 345     =>ruimoku-v6
 346     =jef-china3
 347     =shinjigen
 348     =>cns11643-5
 349     =>cns11643-7
 350     =>gt
 351     =>gt-k
 352     =>daikanwa
 353     =>daikanwa/ho
 354     =>cbeta
 355     =ucs-var-001
 356     =ucs-var-002
 357     =ucs-var-003
 358     =ucs-var-004
 359     =ucs-var-005
 360     =ucs-var-006
 361     =ucs-var-007
 362     =ucs-var-008
 363     =ucs-var-010
 364     =ucs-itaiji-001
 365     =ucs-itaiji-002
 366     =ucs-itaiji-003
 367     =ucs-itaiji-004
 368     =ucs-itaiji-005
 369     =ucs-itaiji-006
 370     =ucs-itaiji-007
 371     =ucs-itaiji-008
 372     =ucs-itaiji-009
 373     =ucs-itaiji-010
 374     =ucs-itaiji-011
 375     =ucs-itaiji-012
 376     =+>ucs-var-003
 377     =>ucs-itaiji-001
 378     =>ucs-itaiji-002
 379     =>ucs-itaiji-003
 380     =>ucs-itaiji-004
 381     =>ucs-itaiji-005
 382     =>ucs-itaiji-006
 383     =>ucs-itaiji-007
 384     =>ucs-itaiji-008
 385     =>ucs-itaiji-009
 386     =>ucs-itaiji-010
 387     =>ucs-itaiji-011
 388     =big5-cdp-var-001
 389     =big5-cdp-var-002
 390     =big5-cdp-var-003
 391     =big5-cdp-var-004
 392     =big5-cdp-var-005
 393     =big5-cdp-var-007
 394     =big5-cdp-var-010
 395     =big5-cdp-itaiji-001
 396     =big5-cdp-itaiji-002
 397     =big5-cdp-itaiji-003
 398     =big5-cdp-itaiji-004
 399     =big5-cdp-itaiji-005
 400     =>big5-cdp-itaiji-001
 401     =>big5-cdp-itaiji-002
 402     =>big5-cdp-itaiji-003
 403     =>big5-cdp-itaiji-004
 404     =>big5-cdp-var-001
 405     =>ucs@iwds-1
 406     =>ucs@cognate
 407     =>ucs@component
 408     =>ucs@iwds-1/normalized
 409     =>ucs-itaiji-001@iwds-1
 410     =>ucs-itaiji-002@iwds-1
 411     =>ucs-itaiji-003@iwds-1
 412     =>ucs-itaiji-006@iwds-1
 413     =>big5-cdp@component
 414     =>big5-cdp@cognate
 415     ==ucs@gb
 416     =ucs@gb
 417     ==ucs-var-002
 418     =ucs@JP/hanazono
 419     =koseki
 420     =shuowen-jiguge))
 421
 422
 423 ;;; @ char-db formatters
 424 ;;;
 425
 426 (defun char-db-make-char-spec (char)
 427   (let (ret char-spec)
 428     (cond ((characterp char)
 429            (cond ((and (setq ret (encode-char char '=ucs 'defined-only))
 430                        (not (and (<= #xE000 ret)(<= ret #xF8FF))))
 431                   (setq char-spec (list (cons '=ucs ret)))
 432                   (cond ((setq ret (get-char-attribute char 'name))
 433                          (setq char-spec (cons (cons 'name ret) char-spec))
 434                          )
 435                         ((setq ret (get-char-attribute char 'name*))
 436                          (setq char-spec (cons (cons 'name* ret) char-spec))
 437                          ))
 438                   )
 439                  ((encode-char char '=mj 'defined-only)
 440                   (setq char-spec nil)
 441                   (dolist (ccs (charset-list))
 442                     (if (and (or (eq ccs '=mj)
 443                                  ;; (eq (charset-property ccs 'iso-ir) 177)
 444                                  (string-match "=ucs@" (symbol-name ccs))
 445                                  )
 446                              (setq ccs (charset-name ccs))
 447                              (null (assq ccs char-spec))
 448                              (setq ret (encode-char char ccs 'defined-only)))
 449                         (setq char-spec (cons (cons ccs ret) char-spec))))
 450                   )
 451                  ((encode-char char '==mj 'defined-only)
 452                   (setq char-spec nil)
 453                   (dolist (ccs (charset-list))
 454                     (if (and (or (eq ccs '==mj)
 455                                  ;; (eq (charset-property ccs 'iso-ir) 177)
 456                                  (string-match "=ucs@" (symbol-name ccs))
 457                                  )
 458                              (setq ccs (charset-name ccs))
 459                              (null (assq ccs char-spec))
 460                              (setq ret (encode-char char ccs 'defined-only)))
 461                         (setq char-spec (cons (cons ccs ret) char-spec))))
 462                   )
 463                  ((encode-char char '=adobe-japan1 'defined-only)
 464                   (setq char-spec nil)
 465                   (dolist (ccs (charset-list))
 466                     (if (and (or (memq ccs
 467                                        '(=adobe-japan1-0
 468                                          =adobe-japan1-1
 469                                          =adobe-japan1-2
 470                                          =adobe-japan1-3
 471                                          =adobe-japan1-4
 472                                          =adobe-japan1-5
 473                                          =adobe-japan1-6
 474                                          ))
 475                                  ;; (eq (charset-property ccs 'iso-ir) 177)
 476                                  (string-match "=ucs@" (symbol-name ccs))
 477                                  )
 478                              (setq ccs (charset-name ccs))
 479                              (null (assq ccs char-spec))
 480                              (setq ret (encode-char char ccs 'defined-only)))
 481                         (setq char-spec (cons (cons ccs ret) char-spec))))
 482                   )
 483                  ((encode-char char '==adobe-japan1 'defined-only)
 484                   (setq char-spec nil)
 485                   (dolist (ccs (charset-list))
 486                     (if (and (or (memq ccs
 487                                        '(==adobe-japan1-0
 488                                          ==adobe-japan1-1
 489                                          ==adobe-japan1-2
 490                                          ==adobe-japan1-3
 491                                          ==adobe-japan1-4
 492                                          ==adobe-japan1-5
 493                                          ==adobe-japan1-6
 494                                          ))
 495                                  ;; (eq (charset-property ccs 'iso-ir) 177)
 496                                  (string-match "=ucs@" (symbol-name ccs))
 497                                  )
 498                              (setq ccs (charset-name ccs))
 499                              (null (assq ccs char-spec))
 500                              (setq ret (encode-char char ccs 'defined-only)))
 501                         (setq char-spec (cons (cons ccs ret) char-spec))))
 502                   )
 503                  ((setq ret
 504                         (catch 'tag
 505                           (let ((rest char-db-coded-charset-priority-list)
 506                                 ccs)
 507                             (while rest
 508                               (setq ccs (charset-name
 509                                          (find-charset (car rest))))
 510                               (if (setq ret
 511                                         (encode-char char ccs
 512                                                      'defined-only))
 513                                   (throw 'tag (cons ccs ret)))
 514                               (setq rest (cdr rest))))))
 515                   (setq char-spec (list ret))
 516                   (dolist (ccs (delq (car ret) (charset-list)))
 517                     (if (and (or (charset-iso-final-char ccs)
 518                                  (memq ccs
 519                                        '(=daikanwa
 520                                          =daikanwa@rev2
 521                                          ;; =gt-k
 522                                          =jis-x0208@1997
 523                                          ))
 524                                  (and (eq (charset-property ccs 'iso-ir) 177)
 525                                       (not (eq (charset-name ccs) '=ucs@big5)))
 526                                  ;; (string-match "=ucs@" (symbol-name ccs))
 527                                  )
 528                              (setq ccs (charset-name ccs))
 529                              (null (assq ccs char-spec))
 530                              (setq ret (encode-char char ccs 'defined-only)))
 531                         (setq char-spec (cons (cons ccs ret) char-spec))))
 532                   (if (null char-spec)
 533                       (setq char-spec (split-char char)))
 534                   (cond ((setq ret (get-char-attribute char 'name))
 535                          (setq char-spec (cons (cons 'name ret) char-spec))
 536                          )
 537                         ((setq ret (get-char-attribute char 'name*))
 538                          (setq char-spec (cons (cons 'name* ret) char-spec))
 539                          ))
 540                   )
 541                  ((setq ret (get-char-attribute
 542                              char 'ideographic-combination))
 543                   (setq char-spec
 544                         (cons (cons 'ideographic-combination ret)
 545                               char-spec))
 546                   ))
 547            char-spec)
 548           ((consp char)
 549            char))))
 550
 551 (defun char-db-insert-char-spec (char &optional readable column
 552                                       required-features)
 553   (unless column
 554     (setq column (current-column)))
 555   (let (char-spec temp-char)
 556     (setq char-spec (char-db-make-char-spec char))
 557     (unless (or (characterp char) ; char
 558                 (condition-case nil
 559                     (setq char (find-char char-spec))
 560                   (error nil)))
 561       ;; define temporary character
 562       ;;   Current implementation is dirty.
 563       (setq temp-char (define-char (cons '(ideograph-daikanwa . 0)
 564                                          char-spec)))
 565       (remove-char-attribute temp-char 'ideograph-daikanwa)
 566       (setq char temp-char))
 567     (insert-char-attributes char
 568                             readable
 569                             (union (mapcar #'car char-spec)
 570                                    required-features)
 571                             nil 'for-sub-node)
 572     (when temp-char
 573       ;; undefine temporary character
 574       ;;   Current implementation is dirty.
 575       (setq char-spec (char-attribute-alist temp-char))
 576       (while char-spec
 577         (remove-char-attribute temp-char (car (car char-spec)))
 578         (setq char-spec (cdr char-spec))))))
 579
 580 (defun char-db-insert-alist (alist &optional readable column)
 581   (unless column
 582     (setq column (current-column)))
 583   (let ((line-breaking
 584          (concat "\n" (make-string (1+ column) ?\ )))
 585         name value
 586         ret al ; cal
 587         key
 588         lbs cell rest separator)
 589     (insert "(")
 590     (while alist
 591       (setq name (car (car alist))
 592             value (cdr (car alist)))
 593       (cond ((eq name 'char)
 594              (insert "(char . ")
 595              (if (setq ret (condition-case nil
 596                                (find-char value)
 597                              (error nil)))
 598                  (progn
 599                    (setq al nil
 600                          ;; cal nil
 601                          )
 602                    (while value
 603                      (setq key (car (car value)))
 604                      ;; (if (find-charset key)
 605                      ;;     (setq cal (cons key cal))
 606                      (setq al (cons key al))
 607                      ;; )
 608                      (setq value (cdr value)))
 609                    (insert-char-attributes ret
 610                                            readable
 611                                            (or al 'none) ; cal
 612                                            nil 'for-sub-node))
 613                (insert (prin1-to-string value)))
 614              (insert ")")
 615              (insert line-breaking))
 616             ((consp value)
 617              (insert (format "(%-18s " name))
 618              (setq lbs (concat "\n" (make-string (current-column) ?\ )))
 619              (while (consp value)
 620                (setq cell (car value))
 621                (if (and (consp cell)
 622                         (consp (car cell))
 623                         (setq ret (condition-case nil
 624                                       (find-char cell)
 625                                     (error nil)))
 626                         )
 627                    (progn
 628                      (setq rest cell
 629                            al nil
 630                            ;; cal nil
 631                            )
 632                      (while rest
 633                        (setq key (car (car rest)))
 634                        ;; (if (find-charset key)
 635                        ;;     (setq cal (cons key cal))
 636                        (setq al (cons key al))
 637                        ;; )
 638                        (setq rest (cdr rest)))
 639                      (if separator
 640                          (insert lbs))
 641                      (insert-char-attributes ret
 642                                              readable
 643                                              al ; cal
 644                                              nil 'for-sub-node)
 645                      (setq separator lbs))
 646                  (if separator
 647                      (insert separator))
 648                  (insert (prin1-to-string cell))
 649                  (setq separator " "))
 650                (setq value (cdr value)))
 651              (insert ")")
 652              (insert line-breaking))
 653             (t
 654              (insert (format "(%-18s . %S)%s"
 655                              name value
 656                              line-breaking))))
 657       (setq alist (cdr alist))))
 658   (insert ")"))
 659
 660 (defun char-db-insert-char-reference (plist &optional readable column)
 661   (unless column
 662     (setq column (current-column)))
 663   (let ((line-breaking
 664          (concat "\n" (make-string (1+ column) ?\ )))
 665         (separator "")
 666         name value)
 667     (insert "(")
 668     (while plist
 669       (setq name (pop plist))
 670       (setq value (pop plist))
 671       (cond ((eq name :char)
 672              (insert separator)
 673              (insert ":char\t")
 674              (cond ((numberp value)
 675                     (setq value (decode-char '=ucs value)))
 676                    ;; ((consp value)
 677                    ;;  (setq value (or (find-char value)
 678                    ;;                  value)))
 679                    )
 680              (char-db-insert-char-spec value readable)
 681              (insert line-breaking)
 682              (setq separator ""))
 683             ((eq name :radical)
 684              (insert (format "%s%s\t%d ; %c%s"
 685                              separator
 686                              name value
 687                              (ideographic-radical value)
 688                              line-breaking))
 689              (setq separator ""))
 690             (t
 691              (insert (format "%s%s\t%S" separator name value))
 692              (setq separator line-breaking)))
 693       ))
 694   (insert ")"))
 695
 696 (defun char-db-decode-isolated-char (ccs code-point)
 697   (let (ret)
 698     (setq ret
 699           (cond ((eq ccs 'arabic-iso8859-6)
 700                  (decode-char ccs code-point))
 701                 ;; ((eq ccs '=gt)
 702                 ;;  (decode-builtin-char '==gt code-point))
 703                 ((and (memq ccs '(=gt-pj-1
 704                                   =gt-pj-2
 705                                   =gt-pj-3
 706                                   =gt-pj-4
 707                                   =gt-pj-5
 708                                   =gt-pj-6
 709                                   =gt-pj-7
 710                                   =gt-pj-8
 711                                   =gt-pj-9
 712                                   =gt-pj-10
 713                                   =gt-pj-11))
 714                       (setq ret (decode-char ccs code-point))
 715                       (setq ret (encode-char ret '=gt 'defined-only)))
 716                  (decode-builtin-char '=gt ret))
 717                 (t
 718                  (decode-builtin-char ccs code-point))))
 719     (cond ((null ret)
 720            (or (decode-char ccs code-point)
 721                (define-char (list (cons ccs code-point)))))
 722           ((and (<= 0 (char-int ret))
 723                 (<= (char-int ret) #x1F))
 724            (decode-char '=ucs (+ #x2400 (char-int ret))))
 725           ((= (char-int ret) #x7F)
 726            ?\u2421)
 727           (t ret))))
 728
 729 (defvar char-db-convert-obsolete-format t)
 730
 731 (defun char-db-insert-ccs-feature (name value line-breaking)
 732   (cond
 733    ((integerp value)
 734     (insert
 735      (format
 736       (cond
 737        ((memq name '(=>iwds-1
 738                      ==shinjigen
 739                      =shinjigen
 740                      =shinjigen@1ed ==shinjigen@1ed
 741                      =shinjigen@rev ==shinjigen@rev
 742                      =shinjigen/+p@rev ==shinjigen/+p@rev
 743                      ===daikanwa/ho ==daikanwa/ho
 744                      =daikanwa/ho =>>daikanwa/ho =>daikanwa/ho))
 745         "(%-18s .  %04d)\t; %c")
 746        ((eq name '=shinjigen@1ed/24pr)
 747         "(%-18s . %04d)\t; %c")
 748        ((or
 749          (memq name
 750                '(===daikanwa
 751                  ==daikanwa =daikanwa =>>daikanwa =>daikanwa
 752                  =daikanwa@rev1 =daikanwa@rev2
 753                  =daikanwa/+p ==daikanwa/+p ===daikanwa/+p
 754                  =>>daikanwa/+p
 755                  =daikanwa/+2p =>>daikanwa/+2p
 756                  =gt ==gt ===gt
 757                  =>>gt =+>gt =>gt
 758                  =gt-k ==gt-k ===gt-k
 759                  =>>gt-k =>gt-k
 760                  =adobe-japan1-0 ==adobe-japan1-0 ===adobe-japan1-0
 761                  =adobe-japan1-1 ==adobe-japan1-1 ===adobe-japan1-1
 762                  =adobe-japan1-2 ==adobe-japan1-2 ===adobe-japan1-2
 763                  =adobe-japan1-3 ==adobe-japan1-3 ===adobe-japan1-3
 764                  =adobe-japan1-4 ==adobe-japan1-4 ===adobe-japan1-4
 765                  =adobe-japan1-5 ==adobe-japan1-5 ===adobe-japan1-5
 766                  =adobe-japan1-6 ==adobe-japan1-6 ===adobe-japan1-6
 767                  =>>adobe-japan1-0 =+>adobe-japan1-0
 768                  =>>adobe-japan1-1 =+>adobe-japan1-1
 769                  =>>adobe-japan1-2 =+>adobe-japan1-2
 770                  =>>adobe-japan1-3 =+>adobe-japan1-3
 771                  =>>adobe-japan1-4 =+>adobe-japan1-4
 772                  =>>adobe-japan1-5 =+>adobe-japan1-5
 773                  =>>adobe-japan1-6 =+>adobe-japan1-6
 774                  =>cbeta =cbeta =>>cbeta ==cbeta ===cbeta
 775                  =zinbun-oracle =>zinbun-oracle
 776                  =daijiten ==daijiten ===daijiten ==>daijiten
 777                  ===hng-jou ===hng-keg ===hng-dng ===hng-mam
 778                  ===hng-drt ===hng-kgk ===hng-myz ===hng-kda
 779                  ===hng-khi ===hng-khm ===hng-fhs ===hng-hok
 780                  ===hng-kyd ===hng-sok
 781                  ===hng-yhk ===hng-kak ===hng-kar ===hng-kae
 782                  ===hng-sys ===hng-tsu ===hng-tzj
 783                  ===hng-hos ===hng-kkh ===hng-nak ===hng-jhk
 784                  ===hng-hod ===hng-gok ===hng-ink ===hng-nto
 785                  ===hng-nkm ===hng-k24 ===hng-ini ===hng-nkk
 786                  ===hng-kcc ===hng-kcj ===hng-kbk ===hng-sik
 787                  ===hng-skk ===hng-kyu ===hng-ksk ===hng-wan
 788                  ===hng-okd ===hng-wad ===hng-kmi
 789                  ===hng-zkd ===hng-doh ===hng-jyu ===hng-tzs
 790                  ===hng-sai ===hng-kad ===hng-kss
 791                  ===hng-kyo ===hng-ykk ===hng-saa
 792                  ===hng-sab ===hng-wks ===hng-wke ===hng-smk
 793                  ===hng-sgs ===hng-sts ===hng-kkd
 794                  =shuowen-jiguge ===shuowen-jiguge4 ===shuowen-jiguge5))
 795          ;; (string-match "^=adobe-" (symbol-name name))
 796          )
 797         "(%-18s . %05d)\t; %c")
 798        ((memq name '(=hanyo-denshi/ks
 799                      ==hanyo-denshi/ks ===hanyo-denshi/ks
 800                      =>>hanyo-denshi/ks
 801                      =koseki ==koseki
 802                      =mj ==mj ===mj =>>mj =>mj
 803                      =zihai mojikyo))
 804         "(%-18s . %06d)\t; %c")
 805        ((memq name '(=hanyo-denshi/tk ==hanyo-denshi/tk))
 806         "(%-18s . %08d)\t; %c")
 807        ((>= (charset-dimension name) 2)
 808         "(%-18s . #x%04X)\t; %c")
 809        (t
 810         "(%-18s . #x%02X)\t; %c"))
 811       name
 812       (if (= (charset-iso-graphic-plane name) 1)
 813           (logior value
 814                   (cond ((= (charset-dimension name) 1)
 815                          #x80)
 816                         ((= (charset-dimension name) 2)
 817                          #x8080)
 818                         ((= (charset-dimension name) 3)
 819                          #x808080)
 820                         (t 0)))
 821         value)
 822       (char-db-decode-isolated-char name value)))
 823     (if (and (= (charset-chars name) 94)
 824              (= (charset-dimension name) 2))
 825         (insert (format " [%02d-%02d]"
 826                         (- (lsh value -8) 32)
 827                         (- (logand value 255) 32))))
 828     )
 829    (t
 830     (insert (format "(%-18s . %s)" name value))
 831     ))
 832   (insert line-breaking))
 833
 834 (defun char-db-insert-relation-feature (char name value line-breaking
 835                                              ccss readable)
 836   (insert (format "(%-18s%s " name line-breaking))
 837   (let ((lbs (concat "\n" (make-string (current-column) ?\ )))
 838         separator cell sources required-features
 839         ret)
 840     (while (consp value)
 841       (setq cell (car value))
 842       (if (integerp cell)
 843           (setq cell (decode-char '=ucs cell)))
 844       (cond
 845        ((eq name '->subsumptive)
 846         (when (or (not (some (lambda (atr)
 847                                (get-char-attribute cell atr))
 848                              char-db-ignored-attributes))
 849                   (some (lambda (ccs)
 850                           (encode-char cell ccs 'defined-only))
 851                         ccss))
 852           (if separator
 853               (insert lbs))
 854           (let ((char-db-ignored-attributes
 855                  (cons '<-subsumptive
 856                        char-db-ignored-attributes)))
 857             (insert-char-attributes cell readable nil nil 'for-sub-node))
 858           (setq separator lbs))
 859         )
 860        ((characterp cell)
 861         (setq sources
 862               (get-char-attribute
 863                char (intern (format "%s*sources" name))))
 864         (setq required-features nil)
 865         (dolist (source sources)
 866           (cond
 867            ((memq source '(JP
 868                            JP/Jouyou
 869                            shinjigen shinjigen@1ed shinjigen@rev))
 870             (setq required-features
 871                   (union required-features
 872                          '(=jis-x0208
 873                            =jis-x0208@1990
 874                            =jis-x0213-1@2000
 875                            =jis-x0213-1@2004
 876                            =jis-x0213-2
 877                            =jis-x0212
 878                            =jis-x0208@1983
 879                            =jis-x0208@1978
 880                            =shinjigen
 881                            =shinjigen@1ed
 882                            =shinjigen@rev
 883                            =shinjigen/+p@rev))))
 884            ((eq source 'CN)
 885             (setq required-features
 886                   (union required-features
 887                          '(=gb2312
 888                            =gb12345
 889                            =iso-ir165)))))
 890           (cond
 891            ((find-charset (setq ret (intern (format "=%s" source))))
 892             (setq required-features
 893                   (cons ret required-features)))
 894            (t (setq required-features
 895                     (cons source required-features)))))
 896         (cond ((string-match "@JP" (symbol-name name))
 897                (setq required-features
 898                      (union required-features
 899                             '(=jis-x0208
 900                               =jis-x0208@1990
 901                               =jis-x0213-1-2000
 902                               =jis-x0213-2-2000
 903                               =jis-x0212
 904                               =jis-x0208@1983
 905                               =jis-x0208@1978))))
 906               ((string-match "@CN" (symbol-name name))
 907                (setq required-features
 908                      (union required-features
 909                             '(=gb2312
 910                               =gb12345
 911                               =iso-ir165)))))
 912         (if separator
 913             (insert lbs))
 914         (if readable
 915             (insert (format "%S" cell))
 916           (char-db-insert-char-spec cell readable
 917                                     nil
 918                                     required-features))
 919         (setq separator lbs))
 920        ((consp cell)
 921         (if separator
 922             (insert lbs))
 923         (if (consp (car cell))
 924             (char-db-insert-char-spec cell readable)
 925           (char-db-insert-char-reference cell readable))
 926         (setq separator lbs))
 927        (t
 928         (if separator
 929             (insert separator))
 930         (insert (prin1-to-string cell))
 931         (setq separator " ")))
 932       (setq value (cdr value)))
 933     (insert ")")
 934     (insert line-breaking)))
 935
 936 (defun insert-char-attributes (char &optional readable attributes column
 937                                     for-sub-node)
 938   (unless column
 939     (setq column (current-column)))
 940   (let (name value ; has-long-ccs-name
 941         rest
 942         radical strokes
 943         (line-breaking
 944          (concat "\n" (make-string (1+ column) ?\ )))
 945         lbs cell separator ret
 946         key al cal
 947         dest-ccss ; sources required-features
 948         ccss)
 949     (let (atr-d)
 950       (setq attributes
 951             (sort (if attributes
 952                       (if (consp attributes)
 953                           (progn
 954                             (dolist (name attributes)
 955                               (unless (memq name char-db-ignored-attributes)
 956                                 (if (find-charset name)
 957                                     (push name ccss))
 958                                 (push name atr-d)))
 959                             atr-d))
 960                     (dolist (name (char-attribute-list))
 961                       (unless (memq name char-db-ignored-attributes)
 962                         (if (find-charset name)
 963                             (push name ccss))
 964                         (push name atr-d)))
 965                     atr-d)
 966                   #'char-attribute-name<)))
 967     (insert "(")
 968     (when (memq '<-subsumptive attributes)
 969       (when (or readable (not for-sub-node))
 970         (when (setq value (get-char-attribute char '<-subsumptive))
 971           (char-db-insert-relation-feature char '<-subsumptive value
 972                                            line-breaking
 973                                            ccss readable)))
 974       (setq attributes (delq '<-subsumptive attributes)))
 975     (when (and (memq '<-denotational attributes)
 976                (setq value (get-char-attribute char '<-denotational)))
 977       (char-db-insert-relation-feature char '<-denotational value
 978                                        line-breaking
 979                                        ccss readable)
 980       (setq attributes (delq '<-denotational attributes)))
 981     (when (and (memq '<-denotational@component attributes)
 982                (setq value (get-char-attribute char '<-denotational@component)))
 983       (char-db-insert-relation-feature char '<-denotational@component value
 984                                        line-breaking
 985                                        ccss readable)
 986       (setq attributes (delq '<-denotational@component attributes)))
 987     (when (and (memq '<-denotational@usage attributes)
 988                (setq value (get-char-attribute char '<-denotational@usage)))
 989       (char-db-insert-relation-feature char '<-denotational@usage value
 990                                        line-breaking
 991                                        ccss readable)
 992       (setq attributes (delq '<-denotational@usage attributes)))
 993     (when (and (memq 'name attributes)
 994                (setq value (get-char-attribute char 'name)))
 995       (insert (format
 996                (if (> (+ (current-column) (length value)) 48)
 997                    "(name . %S)%s"
 998                  "(name               . %S)%s")
 999                value line-breaking))
1000       (setq attributes (delq 'name attributes))
1001       )
1002     (when (and (memq 'name* attributes)
1003                (setq value (get-char-attribute char 'name*)))
1004       (insert (format
1005                (if (> (+ (current-column) (length value)) 48)
1006                    "(name* . %S)%s"
1007                  "(name*              . %S)%s")
1008                value line-breaking))
1009       (setq attributes (delq 'name* attributes))
1010       )
1011     (when (and (memq 'script attributes)
1012                (setq value (get-char-attribute char 'script)))
1013       (insert (format "(script\t\t%s)%s"
1014                       (mapconcat (function prin1-to-string)
1015                                  value " ")
1016                       line-breaking))
1017       (setq attributes (delq 'script attributes))
1018       )
1019     (dolist (name '(=>ucs =>ucs*))
1020       (when (and (memq name attributes)
1021                  (setq value (get-char-attribute char name)))
1022         (insert (format "(%-18s . #x%04X)\t; %c%s"
1023                         name value (decode-char '=ucs value)
1024                         line-breaking))
1025         (setq attributes (delq name attributes))))
1026     (dolist (name '(=>ucs@gb =>ucs@big5))
1027       (when (and (memq name attributes)
1028                  (setq value (get-char-attribute char name)))
1029         (insert (format "(%-18s . #x%04X)\t; %c%s"
1030                         name value
1031                         (decode-char (intern
1032                                       (concat "="
1033                                               (substring
1034                                                (symbol-name name) 2)))
1035                                      value)
1036                         line-breaking))
1037         (setq attributes (delq name attributes))
1038         ))
1039     ;; (dolist (name '(=>daikanwa))
1040     ;;   (when (and (memq name attributes)
1041     ;;              (setq value (get-char-attribute char name)))
1042     ;;     (insert
1043     ;;      (if (integerp value)
1044     ;;          (format "(%-18s . %05d)\t; %c%s"
1045     ;;                  name value (decode-char '=daikanwa value)
1046     ;;                  line-breaking)
1047     ;;        (format "(%-18s %s)\t; %c%s"
1048     ;;                name
1049     ;;                (mapconcat (function prin1-to-string)
1050     ;;                           value " ")
1051     ;;                (char-representative-of-daikanwa char)
1052     ;;                line-breaking)))
1053     ;;     (setq attributes (delq name attributes))))
1054     (when (and (memq 'general-category attributes)
1055                (setq value (get-char-attribute char 'general-category)))
1056       (insert (format
1057                "(general-category\t%s) ; %s%s"
1058                (mapconcat (lambda (cell)
1059                             (format "%S" cell))
1060                           value " ")
1061                (cond ((rassoc value unidata-normative-category-alist)
1062                       "Normative Category")
1063                      ((rassoc value unidata-informative-category-alist)
1064                       "Informative Category")
1065                      (t
1066                       "Unknown Category"))
1067                line-breaking))
1068       (setq attributes (delq 'general-category attributes))
1069       )
1070     (when (and (memq 'bidi-category attributes)
1071                (setq value (get-char-attribute char 'bidi-category)))
1072       (insert (format "(bidi-category\t. %S)%s"
1073                       value
1074                       line-breaking))
1075       (setq attributes (delq 'bidi-category attributes))
1076       )
1077     (unless (or (not (memq 'mirrored attributes))
1078                 (eq (setq value (get-char-attribute char 'mirrored 'empty))
1079                     'empty))
1080       (insert (format "(mirrored\t\t. %S)%s"
1081                       value
1082                       line-breaking))
1083       (setq attributes (delq 'mirrored attributes))
1084       )
1085     (cond
1086      ((and (memq 'decimal-digit-value attributes)
1087            (setq value (get-char-attribute char 'decimal-digit-value)))
1088       (insert (format "(decimal-digit-value . %S)%s"
1089                       value
1090                       line-breaking))
1091       (setq attributes (delq 'decimal-digit-value attributes))
1092       (when (and (memq 'digit-value attributes)
1093                  (setq value (get-char-attribute char 'digit-value)))
1094         (insert (format "(digit-value\t . %S)%s"
1095                         value
1096                         line-breaking))
1097         (setq attributes (delq 'digit-value attributes))
1098         )
1099       (when (and (memq 'numeric-value attributes)
1100                  (setq value (get-char-attribute char 'numeric-value)))
1101         (insert (format "(numeric-value\t . %S)%s"
1102                         value
1103                         line-breaking))
1104         (setq attributes (delq 'numeric-value attributes))
1105         )
1106       )
1107      (t
1108       (when (and (memq 'digit-value attributes)
1109                  (setq value (get-char-attribute char 'digit-value)))
1110         (insert (format "(digit-value\t. %S)%s"
1111                         value
1112                         line-breaking))
1113         (setq attributes (delq 'digit-value attributes))
1114         )
1115       (when (and (memq 'numeric-value attributes)
1116                  (setq value (get-char-attribute char 'numeric-value)))
1117         (insert (format "(numeric-value\t. %S)%s"
1118                         value
1119                         line-breaking))
1120         (setq attributes (delq 'numeric-value attributes))
1121         )))
1122     (when (and (memq 'iso-10646-comment attributes)
1123                (setq value (get-char-attribute char 'iso-10646-comment)))
1124       (insert (format "(iso-10646-comment\t. %S)%s"
1125                       value
1126                       line-breaking))
1127       (setq attributes (delq 'iso-10646-comment attributes))
1128       )
1129     (when (and (memq 'morohashi-daikanwa attributes)
1130                (setq value (get-char-attribute char 'morohashi-daikanwa)))
1131       (insert (format "(morohashi-daikanwa\t%s)%s"
1132                       (mapconcat (function prin1-to-string) value " ")
1133                       line-breaking))
1134       (setq attributes (delq 'morohashi-daikanwa attributes))
1135       )
1136     (setq radical nil
1137           strokes nil)
1138     (when (and (memq 'ideographic-radical attributes)
1139                (setq value (get-char-attribute char 'ideographic-radical)))
1140       (setq radical value)
1141       (insert (format "(ideographic-radical . %S)\t; %c%s"
1142                       radical
1143                       (ideographic-radical radical)
1144                       line-breaking))
1145       (setq attributes (delq 'ideographic-radical attributes))
1146       )
1147     (when (and (memq 'shuowen-radical attributes)
1148                (setq value (get-char-attribute char 'shuowen-radical)))
1149       (insert (format "(shuowen-radical\t. %S)\t; %c%s"
1150                       value
1151                       (shuowen-radical value)
1152                       line-breaking))
1153       (setq attributes (delq 'shuowen-radical attributes))
1154       )
1155     (let (key)
1156       (dolist (domain
1157                (append
1158                 char-db-feature-domains
1159                 (let (dest domain)
1160                   (dolist (feature (char-attribute-list))
1161                     (setq feature (symbol-name feature))
1162                     (when (string-match
1163                            "\\(radical\\|strokes\\)@\\([^@*]+\\)\\(\\*\\|$\\)"
1164                            feature)
1165                       (setq domain (intern (match-string 2 feature)))
1166                      (unless (memq domain dest)
1167                        (setq dest (cons domain dest)))))
1168                   (sort dest #'string<))))
1169         (setq key (intern (format "%s@%s" 'ideographic-radical domain)))
1170         (when (and (memq key attributes)
1171                    (setq value (get-char-attribute char key)))
1172           (setq radical value)
1173           (insert (format "(%s . %S)\t; %c%s"
1174                           key
1175                           radical
1176                           (ideographic-radical radical)
1177                           line-breaking))
1178           (setq attributes (delq key attributes))
1179           )
1180         (setq key (intern (format "%s@%s" 'ideographic-strokes domain)))
1181         (when (and (memq key attributes)
1182                    (setq value (get-char-attribute char key)))
1183           (setq strokes value)
1184           (insert (format "(%s . %S)%s"
1185                           key
1186                           strokes
1187                           line-breaking))
1188           (setq attributes (delq key attributes))
1189           )
1190         (setq key (intern (format "%s@%s" 'total-strokes domain)))
1191         (when (and (memq key attributes)
1192                    (setq value (get-char-attribute char key)))
1193           (insert (format "(%s       . %S)%s"
1194                           key
1195                           value
1196                           line-breaking))
1197           (setq attributes (delq key attributes))
1198           )
1199         (dolist (feature '(ideographic-radical
1200                            ideographic-strokes
1201                            total-strokes))
1202           (setq key (intern (format "%s@%s*sources" feature domain)))
1203           (when (and (memq key attributes)
1204                      (setq value (get-char-attribute char key)))
1205             (insert (format "(%s%s" key line-breaking))
1206             (dolist (cell value)
1207               (insert (format " %s" cell)))
1208             (insert ")")
1209             (insert line-breaking)
1210             (setq attributes (delq key attributes))
1211             ))
1212         ))
1213     (when (and (memq 'ideographic-strokes attributes)
1214                (setq value (get-char-attribute char 'ideographic-strokes)))
1215       (setq strokes value)
1216       (insert (format "(ideographic-strokes . %S)%s"
1217                       strokes
1218                       line-breaking))
1219       (setq attributes (delq 'ideographic-strokes attributes))
1220       )
1221     (when (and (memq 'kangxi-radical attributes)
1222                (setq value (get-char-attribute char 'kangxi-radical)))
1223       (unless (eq value radical)
1224         (insert (format "(kangxi-radical\t . %S)\t; %c%s"
1225                         value
1226                         (ideographic-radical value)
1227                         line-breaking))
1228         (or radical
1229             (setq radical value)))
1230       (setq attributes (delq 'kangxi-radical attributes))
1231       )
1232     (when (and (memq 'kangxi-strokes attributes)
1233                (setq value (get-char-attribute char 'kangxi-strokes)))
1234       (unless (eq value strokes)
1235         (insert (format "(kangxi-strokes\t . %S)%s"
1236                         value
1237                         line-breaking))
1238         (or strokes
1239             (setq strokes value)))
1240       (setq attributes (delq 'kangxi-strokes attributes))
1241       )
1242     ;; (when (and (memq 'japanese-radical attributes)
1243     ;;            (setq value (get-char-attribute char 'japanese-radical)))
1244     ;;   (unless (eq value radical)
1245     ;;     (insert (format "(japanese-radical\t . %S)\t; %c%s"
1246     ;;                     value
1247     ;;                     (ideographic-radical value)
1248     ;;                     line-breaking))
1249     ;;     (or radical
1250     ;;         (setq radical value)))
1251     ;;   (setq attributes (delq 'japanese-radical attributes))
1252     ;;   )
1253     (when (and (memq 'japanese-strokes attributes)
1254                (setq value (get-char-attribute char 'japanese-strokes)))
1255       (unless (eq value strokes)
1256         (insert (format "(japanese-strokes\t . %S)%s"
1257                         value
1258                         line-breaking))
1259         (or strokes
1260             (setq strokes value)))
1261       (setq attributes (delq 'japanese-strokes attributes))
1262       )
1263     (when (and (memq 'cns-radical attributes)
1264                (setq value (get-char-attribute char 'cns-radical)))
1265       (insert (format "(cns-radical\t . %S)\t; %c%s"
1266                       value
1267                       (ideographic-radical value)
1268                       line-breaking))
1269       (setq attributes (delq 'cns-radical attributes))
1270       )
1271     (when (and (memq 'cns-strokes attributes)
1272                (setq value (get-char-attribute char 'cns-strokes)))
1273       (unless (eq value strokes)
1274         (insert (format "(cns-strokes\t . %S)%s"
1275                         value
1276                         line-breaking))
1277         (or strokes
1278             (setq strokes value)))
1279       (setq attributes (delq 'cns-strokes attributes))
1280       )
1281     ;; (when (and (memq 'ideographic- attributes)
1282     ;;            (setq value (get-char-attribute char 'ideographic-)))
1283     ;;   (insert "(ideographic-       ")
1284     ;;   (setq lbs (concat "\n" (make-string (current-column) ?\ ))
1285     ;;         separator nil)
1286     ;;   (while (consp value)
1287     ;;     (setq cell (car value))
1288     ;;     (if (integerp cell)
1289     ;;         (setq cell (decode-char '=ucs cell)))
1290     ;;     (cond ((characterp cell)
1291     ;;            (if separator
1292     ;;                (insert lbs))
1293     ;;            (if readable
1294     ;;                (insert (format "%S" cell))
1295     ;;              (char-db-insert-char-spec cell readable))
1296     ;;            (setq separator lbs))
1297     ;;           ((consp cell)
1298     ;;            (if separator
1299     ;;                (insert lbs))
1300     ;;            (if (consp (car cell))
1301     ;;                (char-db-insert-char-spec cell readable)
1302     ;;              (char-db-insert-char-reference cell readable))
1303     ;;            (setq separator lbs))
1304     ;;           (t
1305     ;;            (if separator
1306     ;;                (insert separator))
1307     ;;            (insert (prin1-to-string cell))
1308     ;;            (setq separator " ")))
1309     ;;     (setq value (cdr value)))
1310     ;;   (insert ")")
1311     ;;   (insert line-breaking)
1312     ;;   (setq attributes (delq 'ideographic- attributes)))
1313     (when (and (memq 'total-strokes attributes)
1314                (setq value (get-char-attribute char 'total-strokes)))
1315       (insert (format "(total-strokes       . %S)%s"
1316                       value
1317                       line-breaking))
1318       (setq attributes (delq 'total-strokes attributes))
1319       )
1320     ;; (when (and (memq '->ideograph attributes)
1321     ;;            (setq value (get-char-attribute char '->ideograph)))
1322     ;;   (insert (format "(->ideograph\t%s)%s"
1323     ;;                   (mapconcat (lambda (code)
1324     ;;                                (cond ((symbolp code)
1325     ;;                                       (symbol-name code))
1326     ;;                                      ((integerp code)
1327     ;;                                       (format "#x%04X" code))
1328     ;;                                      (t
1329     ;;                                       (format "%s %S"
1330     ;;                                               line-breaking code))))
1331     ;;                              value " ")
1332     ;;                   line-breaking))
1333     ;;   (setq attributes (delq '->ideograph attributes))
1334     ;;   )
1335     ;; (when (and (memq '->decomposition attributes)
1336     ;;            (setq value (get-char-attribute char '->decomposition)))
1337     ;;   (insert (format "(->decomposition\t%s)%s"
1338     ;;                   (mapconcat (lambda (code)
1339     ;;                                (cond ((symbolp code)
1340     ;;                                       (symbol-name code))
1341     ;;                                      ((characterp code)
1342     ;;                                       (if readable
1343     ;;                                           (format "%S" code)
1344     ;;                                         (format "#x%04X"
1345     ;;                                                 (char-int code))
1346     ;;                                         ))
1347     ;;                                      ((integerp code)
1348     ;;                                       (format "#x%04X" code))
1349     ;;                                      (t
1350     ;;                                       (format "%s%S" line-breaking code))))
1351     ;;                              value " ")
1352     ;;                   line-breaking))
1353     ;;   (setq attributes (delq '->decomposition attributes))
1354     ;;   )
1355     (if (equal (get-char-attribute char '->titlecase)
1356                (get-char-attribute char '->uppercase))
1357         (setq attributes (delq '->titlecase attributes)))
1358     (when (and (memq '->mojikyo attributes)
1359                (setq value (get-char-attribute char '->mojikyo)))
1360       (insert (format "(->mojikyo\t\t. %06d)\t; %c%s"
1361                       value (decode-char 'mojikyo value)
1362                       line-breaking))
1363       (setq attributes (delq '->mojikyo attributes))
1364       )
1365     (unless readable
1366       (dolist (ignored '(composition
1367                          ->denotational <-subsumptive ->ucs-unified
1368                          ;; ->ideographic-component-forms
1369                          ))
1370         (setq attributes (delq ignored attributes))))
1371     (while attributes
1372       (setq name (car attributes))
1373       (unless (eq (setq value (get-char-attribute char name 'value-is-empty))
1374                   'value-is-empty)
1375         (cond ((setq ret (find-charset name))
1376                (setq name (charset-name ret))
1377                (when (not (memq name dest-ccss))
1378                  (setq dest-ccss (cons name dest-ccss))
1379                  (char-db-insert-ccs-feature name value line-breaking))
1380                )
1381               ((string-match "^=>ucs@" (symbol-name name))
1382                (insert (format "(%-18s . #x%04X)\t; %c%s"
1383                                name value (decode-char '=ucs value)
1384                                line-breaking))
1385                )
1386               ((eq name 'jisx0208-1978/4X)
1387                (insert (format "(%-18s . #x%04X)%s"
1388                                name value
1389                                line-breaking))
1390                )
1391               ((and
1392                 (not readable)
1393                 (not (eq name '->subsumptive))
1394                 (not (eq name '->uppercase))
1395                 (not (eq name '->lowercase))
1396                 (not (eq name '->titlecase))
1397                 (not (eq name '->canonical))
1398                 (not (eq name '->Bopomofo))
1399                 (not (eq name '->mistakable))
1400                 (not (eq name '->ideographic-variants))
1401                 (null (get-char-attribute
1402                        char (intern (format "%s*sources" name))))
1403                 (not (string-match "\\*sources$" (symbol-name name)))
1404                 (null (get-char-attribute
1405                        char (intern (format "%s*note" name))))
1406                 (not (string-match "\\*note$" (symbol-name name)))
1407                 (or (eq name '<-identical)
1408                     (eq name '<-uppercase)
1409                     (eq name '<-lowercase)
1410                     (eq name '<-titlecase)
1411                     (eq name '<-canonical)
1412                     (eq name '<-ideographic-variants)
1413                     ;; (eq name '<-synonyms)
1414                     (string-match "^<-synonyms" (symbol-name name))
1415                     (eq name '<-mistakable)
1416                     (when (string-match "^->" (symbol-name name))
1417                       (cond
1418                        ((string-match "^->fullwidth" (symbol-name name))
1419                         (not (and (consp value)
1420                                   (characterp (car value))
1421                                   (encode-char
1422                                    (car value) '=ucs 'defined-only)))
1423                         )
1424                        (t)))
1425                     ))
1426                )
1427               ((or (eq name 'ideographic-structure)
1428                    (eq name 'ideographic-structure@apparent)
1429                    (eq name 'ideographic-structure@apparent/leftmost)
1430                    (eq name 'ideographic-combination)
1431                    ;; (eq name 'ideographic-)
1432                    (eq name '=decomposition)
1433                    (char-feature-base-name= '=decomposition name)
1434                    (char-feature-base-name= '=>decomposition name)
1435                    ;; (string-match "^=>*decomposition\\(@[^*]+\\)?$"
1436                    ;;               (symbol-name name))
1437                    (string-match "^\\(->\\|<-\\)[^*]*$" (symbol-name name))
1438                    (string-match "^\\(->\\|<-\\)[^*]*\\*sources$"
1439                                  (symbol-name name))
1440                    )
1441                (char-db-insert-relation-feature char name value
1442                                                 line-breaking
1443                                                 ccss readable))
1444               ;; ((memq name '(ideograph=
1445               ;;               original-ideograph-of
1446               ;;               ancient-ideograph-of
1447               ;;               vulgar-ideograph-of
1448               ;;               wrong-ideograph-of
1449               ;;               ;; simplified-ideograph-of
1450               ;;               ideographic-variants
1451               ;;               ;; ideographic-different-form-of
1452               ;;               ))
1453               ;;  (insert (format "(%-18s%s " name line-breaking))
1454               ;;  (setq lbs (concat "\n" (make-string (current-column) ?\ ))
1455               ;;        separator nil)
1456               ;;  (while (consp value)
1457               ;;    (setq cell (car value))
1458               ;;    (if (and (consp cell)
1459               ;;             (consp (car cell)))
1460               ;;        (progn
1461               ;;          (if separator
1462               ;;              (insert lbs))
1463               ;;          (char-db-insert-alist cell readable)
1464               ;;          (setq separator lbs))
1465               ;;      (if separator
1466               ;;          (insert separator))
1467               ;;      (insert (prin1-to-string cell))
1468               ;;      (setq separator " "))
1469               ;;    (setq value (cdr value)))
1470               ;;  (insert ")")
1471               ;;  (insert line-breaking))
1472               ((consp value)
1473                (insert (format "(%-18s " name))
1474                (setq lbs (concat "\n" (make-string (current-column) ?\ ))
1475                      separator nil)
1476                (while (consp value)
1477                  (setq cell (car value))
1478                  (if (and (consp cell)
1479                           (consp (car cell))
1480                           (setq ret (condition-case nil
1481                                         (find-char cell)
1482                                       (error nil))))
1483                      (progn
1484                        (setq rest cell
1485                              al nil
1486                              cal nil)
1487                        (while rest
1488                          (setq key (car (car rest)))
1489                          (if (find-charset key)
1490                              (setq cal (cons key cal))
1491                            (setq al (cons key al)))
1492                          (setq rest (cdr rest)))
1493                        (if separator
1494                            (insert lbs))
1495                        (insert-char-attributes ret
1496                                                readable
1497                                                al ; cal
1498                                                nil 'for-sub-node)
1499                        (setq separator lbs))
1500                    (setq ret (prin1-to-string cell))
1501                    (if separator
1502                        (if (< (+ (current-column)
1503                                  (length ret)
1504                                  (length separator))
1505                               76)
1506                            (insert separator)
1507                          (insert lbs)))
1508                    (insert ret)
1509                    (setq separator " "))
1510                  (setq value (cdr value)))
1511                (insert ")")
1512                (insert line-breaking))
1513               (t
1514                (insert (format "(%-18s" name))
1515                (setq ret (prin1-to-string value))
1516                (unless (< (+ (current-column)
1517                              (length ret)
1518                              3)
1519                           76)
1520                  (insert line-breaking))
1521                (insert " . " ret ")" line-breaking)
1522                ;; (insert (format "(%-18s . %S)%s"
1523                ;;                 name value
1524                ;;                 line-breaking))
1525                )
1526               ))
1527       (setq attributes (cdr attributes)))
1528     (insert ")")))
1529
1530 (defun insert-char-data (char &optional readable
1531                               attributes)
1532   (save-restriction
1533     (narrow-to-region (point)(point))
1534     (insert "(define-char
1535   '")
1536     (insert-char-attributes char readable attributes)
1537     (insert ")\n")
1538     (goto-char (point-min))
1539     (while (re-search-forward "[ \t]+$" nil t)
1540       (replace-match ""))
1541     ;; from tabify.
1542     (goto-char (point-min))
1543     (while (re-search-forward "[ \t][ \t][ \t]*" nil t)
1544       (let ((column (current-column))
1545             (indent-tabs-mode t))
1546         (delete-region (match-beginning 0) (point))
1547         (indent-to column)))
1548     (goto-char (point-max))
1549     ;; (tabify (point-min)(point-max))
1550     ))
1551
1552 (defun insert-char-data-with-variant (char &optional printable
1553                                            no-ucs-unified
1554                                            script excluded-script)
1555   (insert-char-data char printable)
1556   (let ((variants (char-variants char))
1557         rest
1558         variant vs ret)
1559     (setq variants (sort variants #'<))
1560     (setq rest variants)
1561     (setq variants (cons char variants))
1562     (while rest
1563       (setq variant (car rest))
1564       (unless (get-char-attribute variant '<-subsumptive)
1565         (if (and (or (null script)
1566                      (null (setq vs (get-char-attribute variant 'script)))
1567                      (memq script vs))
1568                  (or (null excluded-script)
1569                      (null (setq vs (get-char-attribute variant 'script)))
1570                      (not (memq excluded-script vs))))
1571             (unless (and no-ucs-unified (get-char-attribute variant '=ucs))
1572               (insert-char-data variant printable)
1573               (if (setq ret (char-variants variant))
1574                   (while ret
1575                     (or (memq (car ret) variants)
1576                         ;; (get-char-attribute (car ret) '<-subsumptive)
1577                         (setq rest (nconc rest (list (car ret)))))
1578                     (setq ret (cdr ret)))))))
1579       (setq rest (cdr rest)))))
1580
1581 (defun insert-char-range-data (min max &optional script excluded-script)
1582   (let ((code min)
1583         char)
1584     (while (<= code max)
1585       (setq char (decode-char '=ucs code))
1586       (if (encode-char char '=ucs 'defined-only)
1587           (insert-char-data-with-variant char nil 'no-ucs-unified
1588                                          script excluded-script))
1589       (setq code (1+ code)))))
1590
1591 (defun write-char-range-data-to-file (min max file
1592                                           &optional script excluded-script)
1593   (let ((coding-system-for-write char-db-file-coding-system))
1594     (with-temp-buffer
1595       (insert (format ";; -*- coding: %s -*-\n"
1596                       char-db-file-coding-system))
1597       (insert-char-range-data min max script excluded-script)
1598       (write-region (point-min)(point-max) file))))
1599
1600 (defvar what-character-original-window-configuration)
1601
1602 ;;;###autoload
1603 (defun what-char-definition (char)
1604   (interactive (list (char-after)))
1605   (let ((buf (get-buffer-create "*Character Description*"))
1606         (the-buf (current-buffer))
1607         (win-conf (current-window-configuration)))
1608     (pop-to-buffer buf)
1609     (make-local-variable 'what-character-original-window-configuration)
1610     (setq what-character-original-window-configuration win-conf)
1611     (setq buffer-read-only nil)
1612     (erase-buffer)
1613     (condition-case err
1614         (progn
1615           (insert-char-data-with-variant char 'printable)
1616           (unless (char-attribute-alist char)
1617             (insert (format ";; = %c\n"
1618                             (let* ((rest (split-char char))
1619                                    (ccs (pop rest))
1620                                    (code (pop rest)))
1621                               (while rest
1622                                 (setq code (logior (lsh code 8)
1623                                                    (pop rest))))
1624                               (decode-char ccs code)))))
1625           ;; (char-db-update-comment)
1626           (set-buffer-modified-p nil)
1627           (view-mode the-buf (lambda (buf)
1628                                (set-window-configuration
1629                                 what-character-original-window-configuration)
1630                                ))
1631           (goto-char (point-min)))
1632       (error (progn
1633                (set-window-configuration
1634                 what-character-original-window-configuration)
1635                (signal (car err) (cdr err)))))))
1636
1637
1638 ;;; @ end
1639 ;;;
1640
1641 (provide 'char-db-util)
1642
1643 ;;; char-db-util.el ends here