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

   1 ;;; char-db-util.el --- Character Database utility -*- coding: utf-8-er; -*-
   2
   3 ;; Copyright (C) 1998, 1999, 2000, 2001, 2002, 2003, 2004, 2005, 2006, 2007,
   4 ;;   2008, 2009, 2010, 2011, 2012, 2013, 2014, 2015, 2016, 2017, 2018, 2019,
   5 ;;   2020, 2021, 2022, 2023, 2024 MORIOKA Tomohiko.
   6
   7 ;; Author: MORIOKA Tomohiko <tomo@kanji.zinbun.kyoto-u.ac.jp>
   8 ;; Keywords: CHISE, Character Database, ISO/IEC 10646, UCS, Unicode, MULE.
   9
  10 ;; This file is part of XEmacs CHISE.
  11
  12 ;; XEmacs CHISE is free software; you can redistribute it and/or
  13 ;; modify it under the terms of the GNU General Public License as
  14 ;; published by the Free Software Foundation; either version 2, or (at
  15 ;; your option) any later version.
  16
  17 ;; XEmacs CHISE is distributed in the hope that it will be useful,
  18 ;; but WITHOUT ANY WARRANTY; without even the implied warranty of
  19 ;; MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the GNU
  20 ;; General Public License for more details.
  21
  22 ;; You should have received a copy of the GNU General Public License
  23 ;; along with XEmacs CHISE; see the file COPYING.  If not, write to
  24 ;; the Free Software Foundation, Inc., 59 Temple Place - Suite 330,
  25 ;; Boston, MA 02111-1307, USA.
  26
  27 ;;; Code:
  28
  29 (require 'chise-subr)
  30 (require 'ideograph-subr)
  31
  32 (defconst unidata-normative-category-alist
  33   '(("Lu" letter        uppercase)
  34     ("Ll" letter        lowercase)
  35     ("Lt" letter        titlecase)
  36     ("Mn" mark          non-spacing)
  37     ("Mc" mark          spacing-combining)
  38     ("Me" mark          enclosing)
  39     ("Nd" number        decimal-digit)
  40     ("Nl" number        letter)
  41     ("No" number        other)
  42     ("Zs" separator     space)
  43     ("Zl" separator     line)
  44     ("Zp" separator     paragraph)
  45     ("Cc" other         control)
  46     ("Cf" other         format)
  47     ("Cs" other         surrogate)
  48     ("Co" other         private-use)
  49     ("Cn" other         not-assigned)))
  50
  51 (defconst unidata-informative-category-alist
  52   '(("Lm" letter        modifier)
  53     ("Lo" letter        other)
  54     ("Pc" punctuation   connector)
  55     ("Pd" punctuation   dash)
  56     ("Ps" punctuation   open)
  57     ("Pe" punctuation   close)
  58     ("Pi" punctuation   initial-quote)
  59     ("Pf" punctuation   final-quote)
  60     ("Po" punctuation   other)
  61     ("Sm" symbol        math)
  62     ("Sc" symbol        currency)
  63     ("Sk" symbol        modifier)
  64     ("So" symbol        other)
  65     ))
  66
  67 (defconst shuowen-radicals
  68   [?一 ?上 ?示 ?三 ?王 ?玉 ?玨 ?气 ?士 ?丨 ; 010
  69    ?屮 ?艸 ?蓐 ?茻 ?小 ?八 ?釆 ?半 ?牛 ?犛 ; 020
  70    ?告 ?口 ?凵 ?吅 ?哭 ?走 ?止 ?癶 ?步 ?此 ; 030
  71    ?正 ?是 ?辵 ?彳 ?廴 ?㢟 ?行 ?齒 ?牙 ?足 ; 040
  72    ?疋 ?品 ?龠 ?冊 ?㗊 ?舌 ?干 ?𧮫 ?只 ?㕯 ; 050
  73    ?句 ?丩 ?古 ?十 ?卅 ?言 ?誩 ?音 ?䇂 ?丵 ; 060
  74    ?菐 ?𠬞 ?𠬜 ?共 ?異 ?舁 ?𦥑 ?䢅 ?爨 ?革 ; 070
  75    ?鬲 ?䰜 ?爪 ?𠃨 ?鬥 ?又 ?𠂇 ?史 ?支 ?𦘒 ; 080
  76    ?聿 ?畫 ?隶 ?臤 ?臣 ?殳 ?殺 ?𠘧 ?寸 ?皮 ; 090
  77    ?㼱 ?攴 ?敎 ?卜 ?用 ?爻 ?㸚 ?𥄎 ?目 ?䀠 ; 100
  78    ?眉 ?盾 ?自 ?白 ?鼻 ?皕 ?習 ?羽 ?隹 ?奞 ; 110
  79    ?雈 ?𦫳 ?𥄕 ?羊 ?羴 ?瞿 ?雔 ?雥 ?鳥 ?烏 ; 120
  80    ?𠦒 ?冓 ?幺 ?𢆶 ?叀 ?玄 ?予 ?放 ?𠬪 ?𣦼 ; 130
  81    ?歺 ?死 ?冎 ?骨 ?肉 ?筋 ?刀 ?刃 ?㓞 ?丯 ; 140
  82    ?耒 ?角 ?竹 ?箕 ?丌 ?左 ?工 ?㠭 ?巫 ?甘 ; 150
  83    ?曰 ?乃 ?丂 ?可 ?兮 ?号 ?亏 ?旨 ?喜 ?壴 ; 160
  84 ;  ?旨 ?曰 ?乃 ?丂 ?可 ?兮 ?号 ?亏 ?喜 ?壴 ; 160
  85    ?鼓 ?豈 ?豆 ?豊 ?豐 ?䖒 ?虍 ?虎 ?虤 ?皿 ; 170
  86    ?𠙴 ?去 ?血 ?丶 ?丹 ?青 ?井 ?皀 ?鬯 ?食 ; 180
  87    ?亼 ?會 ?倉 ?入 ?缶 ?矢 ?高 ?冂 ?𩫏 ?京 ; 190
  88    ?亯 ?𣆪 ?畗 ?㐭 ?嗇 ?來 ?麥 ?夊 ?舛 ?䑞 ; 200
  89    ?韋 ?弟 ?夂 ?久 ?桀 ?木 ?東 ?林 ?才 ?叒 ; 210
  90    ?之 ?帀 ?出 ?𣎵 ?生 ?乇 ?𠂹 ?𠌶 ?華 ?𥝌 ; 220
  91    ?稽 ?巢 ?桼 ?束 ?㯻 ?囗 ?員 ?貝 ?邑 ?𨛜 ; 230
  92    ?日 ?旦 ?倝 ?㫃 ?冥 ?晶 ?月 ?有 ?明 ?囧 ; 240
  93    ?夕 ?多 ?毌 ?𢎘 ?𣐺 ?卣 ?齊 ?朿 ?片 ?鼎 ; 250
  94    ?克 ?彔 ?禾 ?秝 ?黍 ?香 ?米 ?毇 ?臼 ?凶 ; 260
  95    ?𣎳 ?𣏟 ?麻 ?尗 ?耑 ?韭 ?瓜 ?瓠 ?宀 ?宮 ; 270
  96    ?呂 ?穴 ?㝱 ?疒 ?冖 ?𠔼 ?冃 ?㒳 ?网 ?襾 ; 280
  97    ?巾 ?巿 ?帛 ?白 ?㡀 ?黹 ?人 ?𠤎 ?匕 ?从 ; 290
  98    ?比 ?北 ?丘 ?㐺 ?𡈼 ?重 ?臥 ?身 ?㐆 ?衣 ; 300
  99    ?裘 ?老 ?毛 ?毳 ?尸 ?尺 ?尾 ?履 ?舟 ?方 ; 310
 100    ?儿 ?兄 ?兂 ?皃 ?𠑹 ?先 ?秃 ?見 ?覞 ?欠 ; 320
 101    ?㱃 ?㳄 ?旡 ?頁 ?𦣻 ?面 ?丏 ?首 ?𥄉 ?須 ; 330
 102    ?彡 ?彣 ?文 ?髟 ?后 ?司 ?卮 ?卩 ?印 ?色 ; 340
 103    ?𠨍 ?辟 ?勹 ?包 ?茍 ?鬼 ?甶 ?厶 ?嵬 ?山 ; 350
 104    ?屾 ?屵 ?广 ?厂 ?丸 ?危 ?石 ?長 ?勿 ?冄 ; 360
 105    ?而 ?豕 ?㣇 ?彑 ?豚 ?豸 ?𤉡 ?易 ?象 ?馬 ; 370
 106    ?𢊁 ?鹿 ?麤 ?㲋 ?兔 ?萈 ?犬 ?㹜 ?鼠 ?能 ; 380
 107    ?熊 ?火 ?炎 ?黑 ?囪 ?焱 ?炙 ?赤 ?大 ?亦 ; 390
 108    ?夨 ?夭 ?交 ?尣 ?壺 ?壹 ?㚔 ?奢 ?亢 ?夲 ; 400
 109    ?夰 ?亣 ?夫 ?立 ?竝 ?囟 ?思 ?心 ?惢 ?水 ; 410
 110    ?沝 ?瀕 ?𡿨 ?巜 ?川 ?泉 ?灥 ?永 ?𠂢 ?谷 ; 420
 111    ?仌 ?雨 ?雲 ?魚 ?𩺰 ?燕 ?龍 ?飛 ?非 ?卂 ; 430
 112    ?𠃉 ?不 ?至 ?西 ?鹵 ?鹽 ?戶 ?門 ?耳 ?𦣝 ; 440
 113    ?手 ?𠦬 ?女 ?毋 ?民 ?丿 ?𠂆 ?乁 ?氏 ?氐 ; 450
 114    ?戈 ?戉 ?我 ?亅 ?琴 ?𠃊 ?亡 ?匸 ?匚 ?曲 ; 460
 115    ?甾 ?瓦 ?弓 ?弜 ?弦 ?系 ?糸 ?素 ?絲 ?率 ; 470
 116    ?虫 ?䖵 ?蟲 ?風 ?它 ?龜 ?黽 ?卵 ?二 ?土 ; 480
 117    ?垚 ?堇 ?里 ?田 ?畕 ?黃 ?男 ?力 ?劦 ?金 ; 490
 118    ?幵 ?勺 ?几 ?且 ?斤 ?斗 ?矛 ?車 ?𠂤 ?𨸏 ; 500
 119    ?𨺅 ?厽 ?四 ?宁 ?叕 ?亞 ?五 ?六 ?七 ?九 ; 510
 120    ?禸 ?嘼 ?甲 ?乙 ?丙 ?丁 ?戊 ?己 ?巴 ?庚 ; 520
 121    ?辛 ?辡 ?壬 ?癸 ?子 ?了 ?孨 ?𠫓 ?丑 ?寅 ; 530
 122    ?卯 ?辰 ?巳 ?午 ?未 ?申 ?酉 ?酋 ?戌 ?亥 ; 540
 123    ])
 124
 125 (defun shuowen-radical (number)
 126   (aref shuowen-radicals (1- number)))
 127
 128 (defvar char-db-file-coding-system 'utf-8-mcs-er)
 129
 130 (defvar char-db-ignored-attributes '(ideographic-products))
 131
 132 (defvar char-db-coded-charset-priority-list
 133   '(ascii
 134     control-1
 135     latin-iso8859-1
 136     latin-iso8859-2
 137     latin-iso8859-3
 138     latin-iso8859-4
 139     latin-iso8859-9
 140     latin-jisx0201
 141     cyrillic-iso8859-5
 142     greek-iso8859-7
 143     thai-tis620
 144     ;; =mj
 145     ;; =adobe-japan1-0
 146     ;; =adobe-japan1-1
 147     ;; =adobe-japan1-2
 148     ;; =adobe-japan1-3
 149     ;; =adobe-japan1-4
 150     ;; =adobe-japan1-5
 151     ;; =adobe-japan1-6
 152     =jis-x0208
 153     =jis-x0208@1978
 154     =jis-x0208@1983
 155     japanese-jisx0212
 156     chinese-gb2312
 157     =jis-x0208@1990
 158     chinese-cns11643-1
 159     chinese-cns11643-2
 160     chinese-cns11643-3
 161     chinese-cns11643-4
 162     chinese-cns11643-5
 163     chinese-cns11643-6
 164     chinese-cns11643-7
 165     =jis-x0213-1
 166     =jis-x0213-1@2000
 167     =jis-x0213-1@2004
 168     =jis-x0213-2
 169     korean-ksc5601
 170     chinese-isoir165
 171     katakana-jisx0201
 172     hebrew-iso8859-8
 173     chinese-gb12345
 174     latin-viscii
 175     ethiopic-ucs
 176     =big5-cdp
 177     =hanyo-denshi/ja
 178     =hanyo-denshi/jb
 179     =hanyo-denshi/jc
 180     =hanyo-denshi/jd
 181     =hanyo-denshi/ft
 182     =hanyo-denshi/ia
 183     =hanyo-denshi/ib
 184     =hanyo-denshi/hg
 185     =hanyo-denshi/ip
 186     =hanyo-denshi/jt
 187     =hanyo-denshi/ks
 188     =hanyo-denshi/tk
 189     ;; ==mj
 190     ;; ==adobe-japan1-0
 191     ;; ==adobe-japan1-1
 192     ;; ==adobe-japan1-2
 193     ;; ==adobe-japan1-3
 194     ;; ==adobe-japan1-4
 195     ;; ==adobe-japan1-5
 196     ;; ==adobe-japan1-6
 197     ==jis-x0208
 198     ==jis-x0213-1
 199     ==jis-x0213-2
 200     ==jis-x0212
 201     ==hanyo-denshi/ja
 202     ==hanyo-denshi/jb
 203     ==hanyo-denshi/jc
 204     ==hanyo-denshi/ft
 205     ==hanyo-denshi/ia
 206     ==hanyo-denshi/ib
 207     ==hanyo-denshi/hg
 208     ==hanyo-denshi/jt
 209     ==hanyo-denshi/ks
 210     ==hanyo-denshi/tk
 211     =gt
 212     =gt-k
 213     =daikanwa
 214     =daijiten
 215     =daikanwa@rev2
 216     =daikanwa@rev1
 217     =daikanwa/+p
 218     =daikanwa/ho
 219     ==daikanwa
 220     ==daijiten
 221     =cbeta
 222     ideograph-hanziku-1
 223     ideograph-hanziku-2
 224     ideograph-hanziku-3
 225     ideograph-hanziku-4
 226     ideograph-hanziku-5
 227     ideograph-hanziku-6
 228     ideograph-hanziku-7
 229     ideograph-hanziku-8
 230     ideograph-hanziku-9
 231     ideograph-hanziku-10
 232     ideograph-hanziku-11
 233     ideograph-hanziku-12
 234     ;; =>>>adobe-japan1-0
 235     ;; =>>>adobe-japan1-1
 236     ;; =>>>adobe-japan1-2
 237     ;; =>>>adobe-japan1-3
 238     ;; =>>>adobe-japan1-4
 239     ;; =>>>adobe-japan1-5
 240     ;; =>>>adobe-japan1-6
 241     ;; =>>>jis-x0208
 242     ;; =>>>jis-x0213-1
 243     ;; =>>>jis-x0213-2
 244     ;; =>>>hanyo-denshi/ja
 245     ;; =>>>hanyo-denshi/jb
 246     ;; =>>>hanyo-denshi/jc
 247     ;; =>>>hanyo-denshi/ft
 248     ;; =>>>hanyo-denshi/ib
 249     ;; =>>>hanyo-denshi/hg
 250     ;; =>>>hanyo-denshi/jt
 251     ;; =>>>hanyo-denshi/ks
 252     ;; =>>>gt
 253     =>>mj
 254     =>>adobe-japan1-0
 255     =>>adobe-japan1-1
 256     =>>adobe-japan1-2
 257     =>>adobe-japan1-3
 258     =>>adobe-japan1-4
 259     =>>adobe-japan1-5
 260     =>>adobe-japan1-6
 261     =>>jis-x0208
 262     =>>jis-x0213-1
 263     =>>jis-x0213-1@2000
 264     =>>jis-x0213-1@2004
 265     =>>jis-x0213-2
 266     =>>jis-x0208@1978
 267     =>>hanyo-denshi/ft
 268     =>>hanyo-denshi/jt
 269     =>>hanyo-denshi/ks
 270     =>>gt
 271     =>>daikanwa
 272     =>>cbeta
 273     =+>jis-x0208
 274     =+>jis-x0213-1
 275     =+>jis-x0213-2
 276     =+>adobe-japan1-0
 277     =+>adobe-japan1-1
 278     =+>adobe-japan1-2
 279     =+>adobe-japan1-3
 280     =+>adobe-japan1-4
 281     =+>adobe-japan1-5
 282     =+>adobe-japan1-6
 283     =+>jis-x0208@1978
 284     =>jis-x0208
 285     =>jis-x0208@1997
 286     =>jis-x0213-1
 287     =>jis-x0213-1@2000
 288     =>jis-x0213-1@2004
 289     =>jis-x0213-2
 290     ==>ucs@bucs
 291     ==>daijiten
 292     ;; =>ucs@hanyo-denshi
 293     =>ucs@iso
 294     =>ucs@unicode
 295     =>ucs@jis
 296     =>ucs@cns
 297     =>ucs@ks
 298     =+>ucs@iso
 299     =+>ucs@unicode
 300     =+>ucs@jis
 301     =+>ucs@jis/1990
 302     =+>ucs@cns
 303     =+>ucs@ks
 304     =>>ucs@iso
 305     =>>ucs@unicode
 306     =>>ucs@jis
 307     =>>ucs@JP
 308     =>>ucs@cns
 309     =>>>ucs@iso
 310     =>>>ucs@unicode
 311     ==ucs@iso
 312     ==ucs@unicode
 313     ;; ==ucs@cns
 314     ==gb2312
 315     ==ks-x1001
 316     ==cns11643-1
 317     ==cns11643-2
 318     ==cns11643-3
 319     ==cns11643-4
 320     ==cns11643-5
 321     ==cns11643-6
 322     ==cns11643-7
 323     ==gt
 324     ==jis-x0208@1990
 325     ;; ==jis-x0208@1983
 326     ==jis-x0208@1978
 327     ==gt-k
 328     =ucs@iso
 329     =ucs@unicode
 330     =ucs@cns
 331     ==big5-cdp
 332     ==cbeta
 333     =>>big5-cdp
 334     =>>gt-k
 335     =+>gt
 336     =+>big5-cdp
 337     =>big5-cdp
 338     =>big5-cdp@iwds-1
 339     =>mj
 340     =>hanyo-denshi/tk
 341     =>iwds-1
 342     =big5
 343     =big5-eten
 344     =zinbun-oracle
 345     =>zinbun-oracle
 346     =ruimoku-v6
 347     =>>ruimoku-v6
 348     ==ruimoku-v6
 349     =>ruimoku-v6
 350     =jef-china3
 351     =shinjigen
 352     =>cns11643-5
 353     =>cns11643-7
 354     =>gt
 355     =>gt-k
 356     =>daikanwa
 357     =>daikanwa/ho
 358     =>cbeta
 359     =ucs-var-001
 360     =ucs-var-002
 361     =ucs-var-003
 362     =ucs-var-004
 363     =ucs-var-005
 364     =ucs-var-006
 365     =ucs-var-007
 366     =ucs-var-008
 367     =ucs-var-010
 368     =ucs-itaiji-001
 369     =ucs-itaiji-002
 370     =ucs-itaiji-003
 371     =ucs-itaiji-004
 372     =ucs-itaiji-005
 373     =ucs-itaiji-006
 374     =ucs-itaiji-007
 375     =ucs-itaiji-008
 376     =ucs-itaiji-009
 377     =ucs-itaiji-010
 378     =ucs-itaiji-011
 379     =ucs-itaiji-012
 380     =ucs-itaiji-013
 381     =ucs-itaiji-014
 382     =ucs-itaiji-015
 383     =ucs-itaiji-016
 384     =ucs-itaiji-018
 385     =+>ucs-var-003
 386     =>ucs-itaiji-001
 387     =>ucs-itaiji-002
 388     =>ucs-itaiji-003
 389     =>ucs-itaiji-004
 390     =>ucs-itaiji-005
 391     =>ucs-itaiji-006
 392     =>ucs-itaiji-007
 393     =>ucs-itaiji-008
 394     =>ucs-itaiji-009
 395     =>ucs-itaiji-010
 396     =>ucs-itaiji-011
 397     =>ucs-itaiji-012
 398     =>ucs-itaiji-014
 399     =>ucs-itaiji-015
 400     =>ucs-itaiji-016
 401     =>ucs-var-001
 402     =big5-cdp-var-001
 403     =big5-cdp-var-002
 404     =big5-cdp-var-003
 405     =big5-cdp-var-004
 406     =big5-cdp-var-005
 407     =big5-cdp-var-007
 408     =big5-cdp-var-008
 409     =big5-cdp-var-010
 410     =big5-cdp-itaiji-001
 411     =big5-cdp-itaiji-002
 412     =big5-cdp-itaiji-003
 413     =big5-cdp-itaiji-004
 414     =big5-cdp-itaiji-005
 415     =>big5-cdp-itaiji-001
 416     =>big5-cdp-itaiji-002
 417     =>big5-cdp-itaiji-003
 418     =>big5-cdp-itaiji-004
 419     =>big5-cdp-itaiji-005
 420     =>big5-cdp-itaiji-006
 421     =>big5-cdp-var-001
 422     =>ucs@iwds-1
 423     =>ucs@cognate
 424     =>ucs@component
 425     =>ucs@iwds-1/normalized
 426     =>ucs-itaiji-001@iwds-1
 427     =>ucs-itaiji-002@iwds-1
 428     =>ucs-itaiji-003@iwds-1
 429     =>ucs-itaiji-006@iwds-1
 430     =>big5-cdp@component
 431     =>big5-cdp@cognate
 432     ==ucs@gb
 433     =ucs@gb
 434     ==ucs-var-002
 435     ==ucs-itaiji-002
 436     =ucs@JP/hanazono
 437     =koseki
 438     =shuowen-jiguge
 439     =>shuowen-jiguge
 440     ===shuowen-jiguge5
 441     ===shuowen-jiguge4
 442     ===chise-hdic-ktb-seal))
 443
 444
 445 ;;; @ char-db formatters
 446 ;;;
 447
 448 (defun char-db-make-char-spec (char)
 449   (let (ret char-spec)
 450     (cond ((characterp char)
 451            (cond ((and (setq ret (encode-char char '=ucs 'defined-only))
 452                        (not (and (<= #xE000 ret)(<= ret #xF8FF))))
 453                   (setq char-spec (list (cons '=ucs ret)))
 454                   (cond ((and (not (and (<= #xE0100 ret)(<= ret #xE01EF)))
 455                               (setq ret (get-char-attribute char 'name)))
 456                          (setq char-spec (cons (cons 'name ret) char-spec))
 457                          )
 458                         ((setq ret (get-char-attribute char 'name*))
 459                          (setq char-spec (cons (cons 'name* ret) char-spec))
 460                          ))
 461                   )
 462                  ((encode-char char '=mj 'defined-only)
 463                   (setq char-spec nil)
 464                   (dolist (ccs (charset-list))
 465                     (if (and (or (eq ccs '=mj)
 466                                  ;; (eq (charset-property ccs 'iso-ir) 177)
 467                                  (string-match "=ucs@" (symbol-name ccs))
 468                                  )
 469                              (setq ccs (charset-name ccs))
 470                              (null (assq ccs char-spec))
 471                              (setq ret (encode-char char ccs 'defined-only)))
 472                         (setq char-spec (cons (cons ccs ret) char-spec))))
 473                   )
 474                  ((encode-char char '==mj 'defined-only)
 475                   (setq char-spec nil)
 476                   (dolist (ccs (charset-list))
 477                     (if (and (or (eq ccs '==mj)
 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                  ((encode-char char '==adobe-japan1 'defined-only)
 507                   (setq char-spec nil)
 508                   (dolist (ccs (charset-list))
 509                     (if (and (or (memq ccs
 510                                        '(==adobe-japan1-0
 511                                          ==adobe-japan1-1
 512                                          ==adobe-japan1-2
 513                                          ==adobe-japan1-3
 514                                          ==adobe-japan1-4
 515                                          ==adobe-japan1-5
 516                                          ==adobe-japan1-6
 517                                          ))
 518                                  ;; (eq (charset-property ccs 'iso-ir) 177)
 519                                  (string-match "=ucs@" (symbol-name ccs))
 520                                  )
 521                              (setq ccs (charset-name ccs))
 522                              (null (assq ccs char-spec))
 523                              (setq ret (encode-char char ccs 'defined-only)))
 524                         (setq char-spec (cons (cons ccs ret) char-spec))))
 525                   )
 526                  ((setq ret
 527                         (catch 'tag
 528                           (let ((rest char-db-coded-charset-priority-list)
 529                                 ccs)
 530                             (while rest
 531                               (setq ccs (charset-name
 532                                          (find-charset (car rest))))
 533                               (if (setq ret
 534                                         (encode-char char ccs
 535                                                      'defined-only))
 536                                   (throw 'tag (cons ccs ret)))
 537                               (setq rest (cdr rest))))))
 538                   (setq char-spec (list ret))
 539                   (dolist (ccs (delq (car ret) (charset-list)))
 540                     (if (and (or (charset-iso-final-char ccs)
 541                                  (memq ccs
 542                                        '(=daikanwa
 543                                          =daikanwa@rev2
 544                                          ;; =gt-k
 545                                          =jis-x0208@1997
 546                                          ))
 547                                  (and (eq (charset-property ccs 'iso-ir) 177)
 548                                       (not (eq (charset-name ccs) '=ucs@big5)))
 549                                  ;; (string-match "=ucs@" (symbol-name ccs))
 550                                  )
 551                              (setq ccs (charset-name ccs))
 552                              (null (assq ccs char-spec))
 553                              (setq ret (encode-char char ccs 'defined-only)))
 554                         (setq char-spec (cons (cons ccs ret) char-spec))))
 555                   (if (null char-spec)
 556                       (setq char-spec (split-char char)))
 557                   (cond ((setq ret (get-char-attribute char 'name))
 558                          (setq char-spec (cons (cons 'name ret) char-spec))
 559                          )
 560                         ((setq ret (get-char-attribute char 'name*))
 561                          (setq char-spec (cons (cons 'name* ret) char-spec))
 562                          ))
 563                   )
 564                  ((setq ret (get-char-attribute
 565                              char 'ideographic-combination))
 566                   (setq char-spec
 567                         (cons (cons 'ideographic-combination ret)
 568                               char-spec))
 569                   ))
 570            char-spec)
 571           ((consp char)
 572            char))))
 573
 574 (defun char-db-insert-char-spec (char &optional readable column
 575                                       required-features)
 576   (unless column
 577     (setq column (current-column)))
 578   (let (char-spec temp-char)
 579     (setq char-spec (char-db-make-char-spec char))
 580     (unless (or (characterp char) ; char
 581                 (condition-case nil
 582                     (setq char (find-char char-spec))
 583                   (error nil)))
 584       ;; define temporary character
 585       ;;   Current implementation is dirty.
 586       (setq temp-char (define-char (cons '(ideograph-daikanwa . 0)
 587                                          char-spec)))
 588       (remove-char-attribute temp-char 'ideograph-daikanwa)
 589       (setq char temp-char))
 590     (insert-char-attributes char
 591                             readable
 592                             (union (mapcar #'car char-spec)
 593                                    required-features)
 594                             nil 'for-sub-node)
 595     (when temp-char
 596       ;; undefine temporary character
 597       ;;   Current implementation is dirty.
 598       (setq char-spec (char-attribute-alist temp-char))
 599       (while char-spec
 600         (remove-char-attribute temp-char (car (car char-spec)))
 601         (setq char-spec (cdr char-spec))))))
 602
 603 (defun char-db-insert-alist (alist &optional readable column)
 604   (unless column
 605     (setq column (current-column)))
 606   (let ((line-breaking
 607          (concat "\n" (make-string (1+ column) ?\ )))
 608         name value
 609         ret al ; cal
 610         key
 611         lbs cell rest separator)
 612     (insert "(")
 613     (while alist
 614       (setq name (car (car alist))
 615             value (cdr (car alist)))
 616       (cond ((eq name 'char)
 617              (insert "(char . ")
 618              (if (setq ret (condition-case nil
 619                                (find-char value)
 620                              (error nil)))
 621                  (progn
 622                    (setq al nil
 623                          ;; cal nil
 624                          )
 625                    (while value
 626                      (setq key (car (car value)))
 627                      ;; (if (find-charset key)
 628                      ;;     (setq cal (cons key cal))
 629                      (setq al (cons key al))
 630                      ;; )
 631                      (setq value (cdr value)))
 632                    (insert-char-attributes ret
 633                                            readable
 634                                            (or al 'none) ; cal
 635                                            nil 'for-sub-node))
 636                (insert (prin1-to-string value)))
 637              (insert ")")
 638              (insert line-breaking))
 639             ((consp value)
 640              (insert (format "(%-18s " name))
 641              (setq lbs (concat "\n" (make-string (current-column) ?\ )))
 642              (while (consp value)
 643                (setq cell (car value))
 644                (if (and (consp cell)
 645                         (consp (car cell))
 646                         (setq ret (condition-case nil
 647                                       (find-char cell)
 648                                     (error nil)))
 649                         )
 650                    (progn
 651                      (setq rest cell
 652                            al nil
 653                            ;; cal nil
 654                            )
 655                      (while rest
 656                        (setq key (car (car rest)))
 657                        ;; (if (find-charset key)
 658                        ;;     (setq cal (cons key cal))
 659                        (setq al (cons key al))
 660                        ;; )
 661                        (setq rest (cdr rest)))
 662                      (if separator
 663                          (insert lbs))
 664                      (insert-char-attributes ret
 665                                              readable
 666                                              al ; cal
 667                                              nil 'for-sub-node)
 668                      (setq separator lbs))
 669                  (if separator
 670                      (insert separator))
 671                  (insert (prin1-to-string cell))
 672                  (setq separator " "))
 673                (setq value (cdr value)))
 674              (insert ")")
 675              (insert line-breaking))
 676             (t
 677              (insert (format "(%-18s . %S)%s"
 678                              name value
 679                              line-breaking))))
 680       (setq alist (cdr alist))))
 681   (insert ")"))
 682
 683 (defun char-db-insert-char-reference (plist &optional readable column)
 684   (unless column
 685     (setq column (current-column)))
 686   (let ((line-breaking
 687          (concat "\n" (make-string (1+ column) ?\ )))
 688         (separator "")
 689         name value)
 690     (insert "(")
 691     (while plist
 692       (setq name (pop plist))
 693       (setq value (pop plist))
 694       (cond ((eq name :char)
 695              (insert separator)
 696              (insert ":char\t")
 697              (cond ((numberp value)
 698                     (setq value (decode-char '=ucs value)))
 699                    ;; ((consp value)
 700                    ;;  (setq value (or (find-char value)
 701                    ;;                  value)))
 702                    )
 703              (char-db-insert-char-spec value readable)
 704              (insert line-breaking)
 705              (setq separator ""))
 706             ((eq name :radical)
 707              (insert (format "%s%s\t%d ; %c%s"
 708                              separator
 709                              name value
 710                              (ideographic-radical value)
 711                              line-breaking))
 712              (setq separator ""))
 713             (t
 714              (insert (format "%s%s\t%S" separator name value))
 715              (setq separator line-breaking)))
 716       ))
 717   (insert ")"))
 718
 719 (defun char-db-decode-isolated-char (ccs code-point)
 720   (let (ret)
 721     (setq ret
 722           (cond ((eq ccs 'arabic-iso8859-6)
 723                  (decode-char ccs code-point))
 724                 ;; ((eq ccs '=gt)
 725                 ;;  (decode-builtin-char '==gt code-point))
 726                 ((and (memq ccs '(=gt-pj-1
 727                                   =gt-pj-2
 728                                   =gt-pj-3
 729                                   =gt-pj-4
 730                                   =gt-pj-5
 731                                   =gt-pj-6
 732                                   =gt-pj-7
 733                                   =gt-pj-8
 734                                   =gt-pj-9
 735                                   =gt-pj-10
 736                                   =gt-pj-11))
 737                       (setq ret (decode-char ccs code-point))
 738                       (setq ret (encode-char ret '=gt 'defined-only)))
 739                  (decode-builtin-char '=gt ret))
 740                 (t
 741                  (decode-builtin-char ccs code-point))))
 742     (cond ((null ret)
 743            (or (decode-char ccs code-point)
 744                (define-char (list (cons ccs code-point)))))
 745           ((and (<= 0 (char-int ret))
 746                 (<= (char-int ret) #x1F))
 747            (decode-char '=ucs (+ #x2400 (char-int ret))))
 748           ((= (char-int ret) #x7F)
 749            ?\u2421)
 750           (t ret))))
 751
 752 (defvar char-db-convert-obsolete-format t)
 753
 754 (defun char-db-insert-ccs-feature (name value line-breaking)
 755   (cond
 756    ((integerp value)
 757     (insert
 758      (format
 759       (cond
 760        ((memq name '(=>iwds-1
 761                      ==shinjigen
 762                      =shinjigen
 763                      =shinjigen@1ed ==shinjigen@1ed
 764                      =shinjigen@rev ==shinjigen@rev
 765                      =shinjigen/+p@rev ==shinjigen/+p@rev
 766                      ===daikanwa/ho ==daikanwa/ho
 767                      =daikanwa/ho =>>daikanwa/ho =>daikanwa/ho))
 768         "(%-18s .  %04d)\t; %c")
 769        ((eq name '=shinjigen@1ed/24pr)
 770         "(%-18s . %04d)\t; %c")
 771        ((or
 772          (memq name
 773                '(===daikanwa
 774                  ==daikanwa =daikanwa =>>daikanwa =>daikanwa
 775                  =daikanwa@rev1 =daikanwa@rev2
 776                  =daikanwa/+p ==daikanwa/+p ===daikanwa/+p
 777                  =>>daikanwa/+p
 778                  =daikanwa/+2p =>>daikanwa/+2p
 779                  =gt ==gt ===gt
 780                  =>>gt =+>gt =>gt
 781                  =gt-k ==gt-k ===gt-k
 782                  =>>gt-k =>gt-k
 783                  =adobe-japan1-0 ==adobe-japan1-0 ===adobe-japan1-0
 784                  =adobe-japan1-1 ==adobe-japan1-1 ===adobe-japan1-1
 785                  =adobe-japan1-2 ==adobe-japan1-2 ===adobe-japan1-2
 786                  =adobe-japan1-3 ==adobe-japan1-3 ===adobe-japan1-3
 787                  =adobe-japan1-4 ==adobe-japan1-4 ===adobe-japan1-4
 788                  =adobe-japan1-5 ==adobe-japan1-5 ===adobe-japan1-5
 789                  =adobe-japan1-6 ==adobe-japan1-6 ===adobe-japan1-6
 790                  =>>adobe-japan1-0 =+>adobe-japan1-0
 791                  =>>adobe-japan1-1 =+>adobe-japan1-1
 792                  =>>adobe-japan1-2 =+>adobe-japan1-2
 793                  =>>adobe-japan1-3 =+>adobe-japan1-3
 794                  =>>adobe-japan1-4 =+>adobe-japan1-4
 795                  =>>adobe-japan1-5 =+>adobe-japan1-5
 796                  =>>adobe-japan1-6 =+>adobe-japan1-6
 797                  =>cbeta =cbeta =>>cbeta ==cbeta ===cbeta
 798                  =zinbun-oracle =>zinbun-oracle
 799                  =daijiten ==daijiten ===daijiten ==>daijiten
 800                  ===hng-jou ===hng-keg ===hng-dng ===hng-mam
 801                  ===hng-drt ===hng-kgk ===hng-myz ===hng-kda
 802                  ===hng-khi ===hng-khm ===hng-fhs ===hng-hok
 803                  ===hng-kyd ===hng-sok
 804                  ===hng-yhk ===hng-kak ===hng-kar ===hng-kae
 805                  ===hng-sys ===hng-tsu ===hng-tzj
 806                  ===hng-hos ===hng-kkh ===hng-nak ===hng-jhk
 807                  ===hng-hod ===hng-gok ===hng-ink ===hng-nto
 808                  ===hng-nkm ===hng-k24 ===hng-ini ===hng-nkk
 809                  ===hng-kcc ===hng-kcj ===hng-kbk ===hng-sik
 810                  ===hng-skk ===hng-kyu ===hng-ksk ===hng-wan
 811                  ===hng-okd ===hng-wad ===hng-kmi
 812                  ===hng-zkd ===hng-doh ===hng-jyu ===hng-tzs
 813                  ===hng-sai ===hng-kad ===hng-kss
 814                  ===hng-kyo ===hng-ykk ===hng-saa
 815                  ===hng-sab ===hng-wks ===hng-wke ===hng-smk
 816                  ===hng-sgs ===hng-sts ===hng-kkd
 817                  =>shuowen-jiguge =shuowen-jiguge
 818                  ===shuowen-jiguge4 ===shuowen-jiguge5 ===shuowen-jiguge-A30))
 819          ;; (string-match "^=adobe-" (symbol-name name))
 820          )
 821         "(%-18s . %05d)\t; %c")
 822        ((memq name '(=hanyo-denshi/ks
 823                      ==hanyo-denshi/ks ===hanyo-denshi/ks
 824                      =>>hanyo-denshi/ks
 825                      =koseki ==koseki
 826                      =mj ==mj ===mj =>>mj =>mj
 827                      =zihai mojikyo))
 828         "(%-18s . %06d)\t; %c")
 829        ((memq name '(=hanyo-denshi/tk =>hanyo-denshi/tk ==hanyo-denshi/tk))
 830         "(%-18s . %08d)\t; %c")
 831        ((>= (charset-dimension name) 2)
 832         "(%-18s . #x%04X)\t; %c")
 833        (t
 834         "(%-18s . #x%02X)\t; %c"))
 835       name
 836       (if (= (charset-iso-graphic-plane name) 1)
 837           (logior value
 838                   (cond ((= (charset-dimension name) 1)
 839                          #x80)
 840                         ((= (charset-dimension name) 2)
 841                          #x8080)
 842                         ((= (charset-dimension name) 3)
 843                          #x808080)
 844                         (t 0)))
 845         value)
 846       (char-db-decode-isolated-char name value)))
 847     (if (and (= (charset-chars name) 94)
 848              (= (charset-dimension name) 2))
 849         (insert (format " [%02d-%02d]"
 850                         (- (lsh value -8) 32)
 851                         (- (logand value 255) 32))))
 852     )
 853    (t
 854     (insert (format "(%-18s . %s)" name value))
 855     ))
 856   (insert line-breaking))
 857
 858 (defun char-db-insert-relation-feature (char name value line-breaking
 859                                              ccss readable)
 860   (insert (format "(%-18s%s " name line-breaking))
 861   (let ((lbs (concat "\n" (make-string (current-column) ?\ )))
 862         separator cell sources required-features
 863         ret)
 864     (while (consp value)
 865       (setq cell (car value))
 866       (if (integerp cell)
 867           (setq cell (decode-char '=ucs cell)))
 868       (cond
 869        ((eq name '->subsumptive)
 870         (when (or (not (some (lambda (atr)
 871                                (get-char-attribute cell atr))
 872                              char-db-ignored-attributes))
 873                   (some (lambda (ccs)
 874                           (encode-char cell ccs 'defined-only))
 875                         ccss))
 876           (if separator
 877               (insert lbs))
 878           (let ((char-db-ignored-attributes
 879                  (cons '<-subsumptive
 880                        char-db-ignored-attributes)))
 881             (insert-char-attributes cell readable nil nil 'for-sub-node))
 882           (setq separator lbs))
 883         )
 884        ((characterp cell)
 885         (setq sources
 886               (get-char-attribute
 887                char (intern (format "%s*sources" name))))
 888         (setq required-features nil)
 889         (dolist (source sources)
 890           (cond
 891            ((memq source '(JP
 892                            JP/Jouyou
 893                            shinjigen shinjigen@1ed shinjigen@rev))
 894             (setq required-features
 895                   (union required-features
 896                          '(=jis-x0208
 897                            =jis-x0208@1990
 898                            =jis-x0213-1@2000
 899                            =jis-x0213-1@2004
 900                            =jis-x0213-2
 901                            =jis-x0212
 902                            =jis-x0208@1983
 903                            =jis-x0208@1978
 904                            =shinjigen
 905                            =shinjigen@1ed
 906                            =shinjigen@rev
 907                            =shinjigen/+p@rev))))
 908            ((eq source 'CN)
 909             (setq required-features
 910                   (union required-features
 911                          '(=gb2312
 912                            =gb12345
 913                            =iso-ir165)))))
 914           (cond
 915            ((find-charset (setq ret (intern (format "=%s" source))))
 916             (setq required-features
 917                   (cons ret required-features)))
 918            (t (setq required-features
 919                     (cons source required-features)))))
 920         (cond ((string-match "@JP" (symbol-name name))
 921                (setq required-features
 922                      (union required-features
 923                             '(=jis-x0208
 924                               =jis-x0208@1990
 925                               =jis-x0213-1-2000
 926                               =jis-x0213-2-2000
 927                               =jis-x0212
 928                               =jis-x0208@1983
 929                               =jis-x0208@1978))))
 930               ((string-match "@CN" (symbol-name name))
 931                (setq required-features
 932                      (union required-features
 933                             '(=gb2312
 934                               =gb12345
 935                               =iso-ir165)))))
 936         (if separator
 937             (insert lbs))
 938         (if readable
 939             (insert (format "%S" cell))
 940           (char-db-insert-char-spec cell readable
 941                                     nil
 942                                     required-features))
 943         (setq separator lbs))
 944        ((consp cell)
 945         (if separator
 946             (insert lbs))
 947         (if (consp (car cell))
 948             (char-db-insert-char-spec cell readable)
 949           (char-db-insert-char-reference cell readable))
 950         (setq separator lbs))
 951        (t
 952         (if separator
 953             (insert separator))
 954         (insert (prin1-to-string cell))
 955         (setq separator " ")))
 956       (setq value (cdr value)))
 957     (insert ")")
 958     (insert line-breaking)))
 959
 960 (defun insert-char-attributes (char &optional readable attributes column
 961                                     for-sub-node)
 962   (unless column
 963     (setq column (current-column)))
 964   (let (name value ; has-long-ccs-name
 965         rest
 966         radical strokes
 967         (line-breaking
 968          (concat "\n" (make-string (1+ column) ?\ )))
 969         lbs cell separator ret
 970         key al cal
 971         dest-ccss ; sources required-features
 972         ccss)
 973     (let (atr-d)
 974       (setq attributes
 975             (sort (if attributes
 976                       (if (consp attributes)
 977                           (progn
 978                             (dolist (name attributes)
 979                               (unless (memq name char-db-ignored-attributes)
 980                                 (if (find-charset name)
 981                                     (push name ccss))
 982                                 (push name atr-d)))
 983                             atr-d))
 984                     (dolist (name (char-attribute-list))
 985                       (unless (memq name char-db-ignored-attributes)
 986                         (if (find-charset name)
 987                             (push name ccss))
 988                         (push name atr-d)))
 989                     atr-d)
 990                   #'char-attribute-name<)))
 991     (insert "(")
 992     (when (memq '<-subsumptive attributes)
 993       (when (or readable (not for-sub-node))
 994         (when (setq value (get-char-attribute char '<-subsumptive))
 995           (char-db-insert-relation-feature char '<-subsumptive value
 996                                            line-breaking
 997                                            ccss readable)))
 998       (setq attributes (delq '<-subsumptive attributes)))
 999     (when (and (memq '<-denotational attributes)
1000                (setq value (get-char-attribute char '<-denotational)))
1001       (char-db-insert-relation-feature char '<-denotational value
1002                                        line-breaking
1003                                        ccss readable)
1004       (setq attributes (delq '<-denotational attributes)))
1005     (when (and (memq '<-denotational@component attributes)
1006                (setq value (get-char-attribute char '<-denotational@component)))
1007       (char-db-insert-relation-feature char '<-denotational@component value
1008                                        line-breaking
1009                                        ccss readable)
1010       (setq attributes (delq '<-denotational@component attributes)))
1011     (when (and (memq '<-denotational@usage attributes)
1012                (setq value (get-char-attribute char '<-denotational@usage)))
1013       (char-db-insert-relation-feature char '<-denotational@usage value
1014                                        line-breaking
1015                                        ccss readable)
1016       (setq attributes (delq '<-denotational@usage attributes)))
1017     (when (and (memq 'name attributes)
1018                (setq value (get-char-attribute char 'name)))
1019       (insert (format
1020                (if (> (+ (current-column) (length value)) 48)
1021                    "(name . %S)%s"
1022                  "(name               . %S)%s")
1023                value line-breaking))
1024       (setq attributes (delq 'name attributes))
1025       )
1026     (when (and (memq 'name* attributes)
1027                (setq value (get-char-attribute char 'name*)))
1028       (insert (format
1029                (if (> (+ (current-column) (length value)) 48)
1030                    "(name* . %S)%s"
1031                  "(name*              . %S)%s")
1032                value line-breaking))
1033       (setq attributes (delq 'name* attributes))
1034       )
1035     (when (and (memq 'script attributes)
1036                (setq value (get-char-attribute char 'script)))
1037       (insert (format "(script\t\t%s)%s"
1038                       (mapconcat (function prin1-to-string)
1039                                  value " ")
1040                       line-breaking))
1041       (setq attributes (delq 'script attributes))
1042       )
1043     (dolist (name '(=>ucs =>ucs*))
1044       (when (and (memq name attributes)
1045                  (setq value (get-char-attribute char name)))
1046         (insert (format "(%-18s . #x%04X)\t; %c%s"
1047                         name value (decode-char '=ucs value)
1048                         line-breaking))
1049         (setq attributes (delq name attributes))))
1050     (dolist (name '(=>ucs@gb =>ucs@big5))
1051       (when (and (memq name attributes)
1052                  (setq value (get-char-attribute char name)))
1053         (insert (format "(%-18s . #x%04X)\t; %c%s"
1054                         name value
1055                         (decode-char (intern
1056                                       (concat "="
1057                                               (substring
1058                                                (symbol-name name) 2)))
1059                                      value)
1060                         line-breaking))
1061         (setq attributes (delq name attributes))
1062         ))
1063     ;; (dolist (name '(=>daikanwa))
1064     ;;   (when (and (memq name attributes)
1065     ;;              (setq value (get-char-attribute char name)))
1066     ;;     (insert
1067     ;;      (if (integerp value)
1068     ;;          (format "(%-18s . %05d)\t; %c%s"
1069     ;;                  name value (decode-char '=daikanwa value)
1070     ;;                  line-breaking)
1071     ;;        (format "(%-18s %s)\t; %c%s"
1072     ;;                name
1073     ;;                (mapconcat (function prin1-to-string)
1074     ;;                           value " ")
1075     ;;                (char-representative-of-daikanwa char)
1076     ;;                line-breaking)))
1077     ;;     (setq attributes (delq name attributes))))
1078     (when (and (memq 'general-category attributes)
1079                (setq value (get-char-attribute char 'general-category)))
1080       (insert (format
1081                "(general-category\t%s) ; %s%s"
1082                (mapconcat (lambda (cell)
1083                             (format "%S" cell))
1084                           value " ")
1085                (cond ((rassoc value unidata-normative-category-alist)
1086                       "Normative Category")
1087                      ((rassoc value unidata-informative-category-alist)
1088                       "Informative Category")
1089                      (t
1090                       "Unknown Category"))
1091                line-breaking))
1092       (setq attributes (delq 'general-category attributes))
1093       )
1094     (when (and (memq 'bidi-category attributes)
1095                (setq value (get-char-attribute char 'bidi-category)))
1096       (insert (format "(bidi-category\t. %S)%s"
1097                       value
1098                       line-breaking))
1099       (setq attributes (delq 'bidi-category attributes))
1100       )
1101     (unless (or (not (memq 'mirrored attributes))
1102                 (eq (setq value (get-char-attribute char 'mirrored 'empty))
1103                     'empty))
1104       (insert (format "(mirrored\t\t. %S)%s"
1105                       value
1106                       line-breaking))
1107       (setq attributes (delq 'mirrored attributes))
1108       )
1109     (cond
1110      ((and (memq 'decimal-digit-value attributes)
1111            (setq value (get-char-attribute char 'decimal-digit-value)))
1112       (insert (format "(decimal-digit-value . %S)%s"
1113                       value
1114                       line-breaking))
1115       (setq attributes (delq 'decimal-digit-value attributes))
1116       (when (and (memq 'digit-value attributes)
1117                  (setq value (get-char-attribute char 'digit-value)))
1118         (insert (format "(digit-value\t . %S)%s"
1119                         value
1120                         line-breaking))
1121         (setq attributes (delq 'digit-value attributes))
1122         )
1123       (when (and (memq 'numeric-value attributes)
1124                  (setq value (get-char-attribute char 'numeric-value)))
1125         (insert (format "(numeric-value\t . %S)%s"
1126                         value
1127                         line-breaking))
1128         (setq attributes (delq 'numeric-value attributes))
1129         )
1130       )
1131      (t
1132       (when (and (memq 'digit-value attributes)
1133                  (setq value (get-char-attribute char 'digit-value)))
1134         (insert (format "(digit-value\t. %S)%s"
1135                         value
1136                         line-breaking))
1137         (setq attributes (delq 'digit-value attributes))
1138         )
1139       (when (and (memq 'numeric-value attributes)
1140                  (setq value (get-char-attribute char 'numeric-value)))
1141         (insert (format "(numeric-value\t. %S)%s"
1142                         value
1143                         line-breaking))
1144         (setq attributes (delq 'numeric-value attributes))
1145         )))
1146     (when (and (memq 'iso-10646-comment attributes)
1147                (setq value (get-char-attribute char 'iso-10646-comment)))
1148       (insert (format "(iso-10646-comment\t. %S)%s"
1149                       value
1150                       line-breaking))
1151       (setq attributes (delq 'iso-10646-comment attributes))
1152       )
1153     (when (and (memq 'morohashi-daikanwa attributes)
1154                (setq value (get-char-attribute char 'morohashi-daikanwa)))
1155       (insert (format "(morohashi-daikanwa\t%s)%s"
1156                       (mapconcat (function prin1-to-string) value " ")
1157                       line-breaking))
1158       (setq attributes (delq 'morohashi-daikanwa attributes))
1159       )
1160     (setq radical nil
1161           strokes nil)
1162     (when (and (memq 'ideographic-radical attributes)
1163                (setq value (get-char-attribute char 'ideographic-radical)))
1164       (setq radical value)
1165       (insert (format "(ideographic-radical . %S)\t; %c%s"
1166                       radical
1167                       (ideographic-radical radical)
1168                       line-breaking))
1169       (setq attributes (delq 'ideographic-radical attributes))
1170       )
1171     (when (and (memq 'shuowen-radical attributes)
1172                (setq value (get-char-attribute char 'shuowen-radical)))
1173       (insert (format "(shuowen-radical\t. %S)\t; %c%s"
1174                       value
1175                       (shuowen-radical value)
1176                       line-breaking))
1177       (setq attributes (delq 'shuowen-radical attributes))
1178       )
1179     (let (key)
1180       (dolist (domain
1181                (append
1182                 char-db-feature-domains
1183                 (let (dest domain)
1184                   (dolist (feature (char-attribute-list))
1185                     (setq feature (symbol-name feature))
1186                     (when (string-match
1187                            "\\(radical\\|strokes\\)@\\([^@*]+\\)\\(\\*\\|$\\)"
1188                            feature)
1189                       (setq domain (intern (match-string 2 feature)))
1190                      (unless (memq domain dest)
1191                        (setq dest (cons domain dest)))))
1192                   (sort dest #'string<))))
1193         (setq key (intern (format "%s@%s" 'ideographic-radical domain)))
1194         (when (and (memq key attributes)
1195                    (setq value (get-char-attribute char key)))
1196           (setq radical value)
1197           (insert (format "(%s . %S)\t; %c%s"
1198                           key
1199                           radical
1200                           (ideographic-radical radical)
1201                           line-breaking))
1202           (setq attributes (delq key attributes))
1203           )
1204         (setq key (intern (format "%s@%s" 'ideographic-strokes domain)))
1205         (when (and (memq key attributes)
1206                    (setq value (get-char-attribute char key)))
1207           (setq strokes value)
1208           (insert (format "(%s . %S)%s"
1209                           key
1210                           strokes
1211                           line-breaking))
1212           (setq attributes (delq key attributes))
1213           )
1214         (setq key (intern (format "%s@%s" 'total-strokes domain)))
1215         (when (and (memq key attributes)
1216                    (setq value (get-char-attribute char key)))
1217           (insert (format "(%s       . %S)%s"
1218                           key
1219                           value
1220                           line-breaking))
1221           (setq attributes (delq key attributes))
1222           )
1223         (dolist (feature '(ideographic-radical
1224                            ideographic-strokes
1225                            total-strokes))
1226           (setq key (intern (format "%s@%s*sources" feature domain)))
1227           (when (and (memq key attributes)
1228                      (setq value (get-char-attribute char key)))
1229             (insert (format "(%s%s" key line-breaking))
1230             (dolist (cell value)
1231               (insert (format " %s" cell)))
1232             (insert ")")
1233             (insert line-breaking)
1234             (setq attributes (delq key attributes))
1235             ))
1236         ))
1237     (when (and (memq 'ideographic-strokes attributes)
1238                (setq value (get-char-attribute char 'ideographic-strokes)))
1239       (setq strokes value)
1240       (insert (format "(ideographic-strokes . %S)%s"
1241                       strokes
1242                       line-breaking))
1243       (setq attributes (delq 'ideographic-strokes attributes))
1244       )
1245     (when (and (memq 'kangxi-radical attributes)
1246                (setq value (get-char-attribute char 'kangxi-radical)))
1247       (unless (eq value radical)
1248         (insert (format "(kangxi-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 'kangxi-radical attributes))
1255       )
1256     (when (and (memq 'kangxi-strokes attributes)
1257                (setq value (get-char-attribute char 'kangxi-strokes)))
1258       (unless (eq value strokes)
1259         (insert (format "(kangxi-strokes\t . %S)%s"
1260                         value
1261                         line-breaking))
1262         (or strokes
1263             (setq strokes value)))
1264       (setq attributes (delq 'kangxi-strokes attributes))
1265       )
1266     ;; (when (and (memq 'japanese-radical attributes)
1267     ;;            (setq value (get-char-attribute char 'japanese-radical)))
1268     ;;   (unless (eq value radical)
1269     ;;     (insert (format "(japanese-radical\t . %S)\t; %c%s"
1270     ;;                     value
1271     ;;                     (ideographic-radical value)
1272     ;;                     line-breaking))
1273     ;;     (or radical
1274     ;;         (setq radical value)))
1275     ;;   (setq attributes (delq 'japanese-radical attributes))
1276     ;;   )
1277     (when (and (memq 'japanese-strokes attributes)
1278                (setq value (get-char-attribute char 'japanese-strokes)))
1279       (unless (eq value strokes)
1280         (insert (format "(japanese-strokes\t . %S)%s"
1281                         value
1282                         line-breaking))
1283         (or strokes
1284             (setq strokes value)))
1285       (setq attributes (delq 'japanese-strokes attributes))
1286       )
1287     (when (and (memq 'cns-radical attributes)
1288                (setq value (get-char-attribute char 'cns-radical)))
1289       (insert (format "(cns-radical\t . %S)\t; %c%s"
1290                       value
1291                       (ideographic-radical value)
1292                       line-breaking))
1293       (setq attributes (delq 'cns-radical attributes))
1294       )
1295     (when (and (memq 'cns-strokes attributes)
1296                (setq value (get-char-attribute char 'cns-strokes)))
1297       (unless (eq value strokes)
1298         (insert (format "(cns-strokes\t . %S)%s"
1299                         value
1300                         line-breaking))
1301         (or strokes
1302             (setq strokes value)))
1303       (setq attributes (delq 'cns-strokes attributes))
1304       )
1305     ;; (when (and (memq 'ideographic- attributes)
1306     ;;            (setq value (get-char-attribute char 'ideographic-)))
1307     ;;   (insert "(ideographic-       ")
1308     ;;   (setq lbs (concat "\n" (make-string (current-column) ?\ ))
1309     ;;         separator nil)
1310     ;;   (while (consp value)
1311     ;;     (setq cell (car value))
1312     ;;     (if (integerp cell)
1313     ;;         (setq cell (decode-char '=ucs cell)))
1314     ;;     (cond ((characterp cell)
1315     ;;            (if separator
1316     ;;                (insert lbs))
1317     ;;            (if readable
1318     ;;                (insert (format "%S" cell))
1319     ;;              (char-db-insert-char-spec cell readable))
1320     ;;            (setq separator lbs))
1321     ;;           ((consp cell)
1322     ;;            (if separator
1323     ;;                (insert lbs))
1324     ;;            (if (consp (car cell))
1325     ;;                (char-db-insert-char-spec cell readable)
1326     ;;              (char-db-insert-char-reference cell readable))
1327     ;;            (setq separator lbs))
1328     ;;           (t
1329     ;;            (if separator
1330     ;;                (insert separator))
1331     ;;            (insert (prin1-to-string cell))
1332     ;;            (setq separator " ")))
1333     ;;     (setq value (cdr value)))
1334     ;;   (insert ")")
1335     ;;   (insert line-breaking)
1336     ;;   (setq attributes (delq 'ideographic- attributes)))
1337     (when (and (memq 'total-strokes attributes)
1338                (setq value (get-char-attribute char 'total-strokes)))
1339       (insert (format "(total-strokes       . %S)%s"
1340                       value
1341                       line-breaking))
1342       (setq attributes (delq 'total-strokes attributes))
1343       )
1344     ;; (when (and (memq '->ideograph attributes)
1345     ;;            (setq value (get-char-attribute char '->ideograph)))
1346     ;;   (insert (format "(->ideograph\t%s)%s"
1347     ;;                   (mapconcat (lambda (code)
1348     ;;                                (cond ((symbolp code)
1349     ;;                                       (symbol-name code))
1350     ;;                                      ((integerp code)
1351     ;;                                       (format "#x%04X" code))
1352     ;;                                      (t
1353     ;;                                       (format "%s %S"
1354     ;;                                               line-breaking code))))
1355     ;;                              value " ")
1356     ;;                   line-breaking))
1357     ;;   (setq attributes (delq '->ideograph attributes))
1358     ;;   )
1359     ;; (when (and (memq '->decomposition attributes)
1360     ;;            (setq value (get-char-attribute char '->decomposition)))
1361     ;;   (insert (format "(->decomposition\t%s)%s"
1362     ;;                   (mapconcat (lambda (code)
1363     ;;                                (cond ((symbolp code)
1364     ;;                                       (symbol-name code))
1365     ;;                                      ((characterp code)
1366     ;;                                       (if readable
1367     ;;                                           (format "%S" code)
1368     ;;                                         (format "#x%04X"
1369     ;;                                                 (char-int code))
1370     ;;                                         ))
1371     ;;                                      ((integerp code)
1372     ;;                                       (format "#x%04X" code))
1373     ;;                                      (t
1374     ;;                                       (format "%s%S" line-breaking code))))
1375     ;;                              value " ")
1376     ;;                   line-breaking))
1377     ;;   (setq attributes (delq '->decomposition attributes))
1378     ;;   )
1379     (if (equal (get-char-attribute char '->titlecase)
1380                (get-char-attribute char '->uppercase))
1381         (setq attributes (delq '->titlecase attributes)))
1382     (when (and (memq '->mojikyo attributes)
1383                (setq value (get-char-attribute char '->mojikyo)))
1384       (insert (format "(->mojikyo\t\t. %06d)\t; %c%s"
1385                       value (decode-char 'mojikyo value)
1386                       line-breaking))
1387       (setq attributes (delq '->mojikyo attributes))
1388       )
1389     (unless readable
1390       (dolist (ignored '(composition
1391                          ->denotational <-subsumptive ->ucs-unified
1392                          ;; ->ideographic-component-forms
1393                          ))
1394         (setq attributes (delq ignored attributes))))
1395     (while attributes
1396       (setq name (car attributes))
1397       (unless (eq (setq value (get-char-attribute char name 'value-is-empty))
1398                   'value-is-empty)
1399         (cond ((setq ret (find-charset name))
1400                (setq name (charset-name ret))
1401                (when (not (memq name dest-ccss))
1402                  (setq dest-ccss (cons name dest-ccss))
1403                  (char-db-insert-ccs-feature name value line-breaking))
1404                )
1405               ((string-match "^=>ucs@" (symbol-name name))
1406                (insert (format "(%-18s . #x%04X)\t; %c%s"
1407                                name value (decode-char '=ucs value)
1408                                line-breaking))
1409                )
1410               ((eq name 'jisx0208-1978/4X)
1411                (insert (format "(%-18s . #x%04X)%s"
1412                                name value
1413                                line-breaking))
1414                )
1415               ((and
1416                 (not readable)
1417                 (not (eq name '->subsumptive))
1418                 (not (eq name '->uppercase))
1419                 (not (eq name '->lowercase))
1420                 (not (eq name '->titlecase))
1421                 (not (eq name '->canonical))
1422                 (not (eq name '->Bopomofo))
1423                 (not (eq name '->mistakable))
1424                 (not (eq name '->ideographic-variants))
1425                 (null (get-char-attribute
1426                        char (intern (format "%s*sources" name))))
1427                 (not (string-match "\\*sources$" (symbol-name name)))
1428                 (null (get-char-attribute
1429                        char (intern (format "%s*note" name))))
1430                 (not (string-match "\\*note$" (symbol-name name)))
1431                 (or (eq name '<-identical)
1432                     (eq name '<-uppercase)
1433                     (eq name '<-lowercase)
1434                     (eq name '<-titlecase)
1435                     (eq name '<-canonical)
1436                     (eq name '<-ideographic-variants)
1437                     ;; (eq name '<-synonyms)
1438                     (string-match "^<-synonyms" (symbol-name name))
1439                     (eq name '<-mistakable)
1440                     (when (string-match "^->" (symbol-name name))
1441                       (cond
1442                        ((string-match "^->fullwidth" (symbol-name name))
1443                         (not (and (consp value)
1444                                   (characterp (car value))
1445                                   (encode-char
1446                                    (car value) '=ucs 'defined-only)))
1447                         )
1448                        (t)))
1449                     ))
1450                )
1451               ((or (eq name 'ideographic-structure)
1452                    (eq name 'ideographic-structure@apparent)
1453                    (eq name 'ideographic-structure@apparent/leftmost)
1454                    (eq name 'ideographic-structure@apparent/rightmost)
1455                    (eq name 'ideographic-combination)
1456                    ;; (eq name 'ideographic-)
1457                    (eq name '=decomposition)
1458                    (char-feature-base-name= '=decomposition name)
1459                    (char-feature-base-name= '=>decomposition name)
1460                    ;; (string-match "^=>*decomposition\\(@[^*]+\\)?$"
1461                    ;;               (symbol-name name))
1462                    (string-match "^\\(->\\|<-\\)[^*]*$" (symbol-name name))
1463                    (string-match "^\\(->\\|<-\\)[^*]*\\*sources$"
1464                                  (symbol-name name))
1465                    )
1466                (char-db-insert-relation-feature char name value
1467                                                 line-breaking
1468                                                 ccss readable))
1469               ;; ((memq name '(ideograph=
1470               ;;               original-ideograph-of
1471               ;;               ancient-ideograph-of
1472               ;;               vulgar-ideograph-of
1473               ;;               wrong-ideograph-of
1474               ;;               ;; simplified-ideograph-of
1475               ;;               ideographic-variants
1476               ;;               ;; ideographic-different-form-of
1477               ;;               ))
1478               ;;  (insert (format "(%-18s%s " name line-breaking))
1479               ;;  (setq lbs (concat "\n" (make-string (current-column) ?\ ))
1480               ;;        separator nil)
1481               ;;  (while (consp value)
1482               ;;    (setq cell (car value))
1483               ;;    (if (and (consp cell)
1484               ;;             (consp (car cell)))
1485               ;;        (progn
1486               ;;          (if separator
1487               ;;              (insert lbs))
1488               ;;          (char-db-insert-alist cell readable)
1489               ;;          (setq separator lbs))
1490               ;;      (if separator
1491               ;;          (insert separator))
1492               ;;      (insert (prin1-to-string cell))
1493               ;;      (setq separator " "))
1494               ;;    (setq value (cdr value)))
1495               ;;  (insert ")")
1496               ;;  (insert line-breaking))
1497               ((consp value)
1498                (insert (format "(%-18s " name))
1499                (setq lbs (concat "\n" (make-string (current-column) ?\ ))
1500                      separator nil)
1501                (while (consp value)
1502                  (setq cell (car value))
1503                  (if (and (consp cell)
1504                           (consp (car cell))
1505                           (setq ret (condition-case nil
1506                                         (find-char cell)
1507                                       (error nil))))
1508                      (progn
1509                        (setq rest cell
1510                              al nil
1511                              cal nil)
1512                        (while rest
1513                          (setq key (car (car rest)))
1514                          (if (find-charset key)
1515                              (setq cal (cons key cal))
1516                            (setq al (cons key al)))
1517                          (setq rest (cdr rest)))
1518                        (if separator
1519                            (insert lbs))
1520                        (insert-char-attributes ret
1521                                                readable
1522                                                al ; cal
1523                                                nil 'for-sub-node)
1524                        (setq separator lbs))
1525                    (setq ret (prin1-to-string cell))
1526                    (if separator
1527                        (if (< (+ (current-column)
1528                                  (length ret)
1529                                  (length separator))
1530                               76)
1531                            (insert separator)
1532                          (insert lbs)))
1533                    (insert ret)
1534                    (setq separator " "))
1535                  (setq value (cdr value)))
1536                (insert ")")
1537                (insert line-breaking))
1538               (t
1539                (insert (format "(%-18s" name))
1540                (setq ret (prin1-to-string value))
1541                (unless (< (+ (current-column)
1542                              (length ret)
1543                              3)
1544                           76)
1545                  (insert line-breaking))
1546                (insert " . " ret ")" line-breaking)
1547                ;; (insert (format "(%-18s . %S)%s"
1548                ;;                 name value
1549                ;;                 line-breaking))
1550                )
1551               ))
1552       (setq attributes (cdr attributes)))
1553     (insert ")")))
1554
1555 (defun insert-char-data (char &optional readable
1556                               attributes)
1557   (save-restriction
1558     (narrow-to-region (point)(point))
1559     (insert "(define-char
1560   '")
1561     (insert-char-attributes char readable attributes)
1562     (insert ")\n")
1563     (goto-char (point-min))
1564     (while (re-search-forward "[ \t]+$" nil t)
1565       (replace-match ""))
1566     ;; from tabify.
1567     (goto-char (point-min))
1568     (while (re-search-forward "[ \t][ \t][ \t]*" nil t)
1569       (let ((column (current-column))
1570             (indent-tabs-mode t))
1571         (delete-region (match-beginning 0) (point))
1572         (indent-to column)))
1573     (goto-char (point-max))
1574     ;; (tabify (point-min)(point-max))
1575     ))
1576
1577 (defun insert-char-data-with-variant (char &optional printable
1578                                            no-ucs-unified
1579                                            script excluded-script)
1580   (insert-char-data char printable)
1581   (let ((variants (char-variants char))
1582         rest
1583         variant vs ret)
1584     (setq variants (sort variants #'<))
1585     (setq rest variants)
1586     (setq variants (cons char variants))
1587     (while rest
1588       (setq variant (car rest))
1589       (unless (get-char-attribute variant '<-subsumptive)
1590         (if (and (or (null script)
1591                      (null (setq vs (get-char-attribute variant 'script)))
1592                      (memq script vs))
1593                  (or (null excluded-script)
1594                      (null (setq vs (get-char-attribute variant 'script)))
1595                      (not (memq excluded-script vs))))
1596             (unless (and no-ucs-unified (get-char-attribute variant '=ucs))
1597               (insert-char-data variant printable)
1598               (if (setq ret (char-variants variant))
1599                   (while ret
1600                     (or (memq (car ret) variants)
1601                         ;; (get-char-attribute (car ret) '<-subsumptive)
1602                         (setq rest (nconc rest (list (car ret)))))
1603                     (setq ret (cdr ret)))))))
1604       (setq rest (cdr rest)))))
1605
1606 (defun insert-char-range-data (min max &optional script excluded-script)
1607   (let ((code min)
1608         char)
1609     (while (<= code max)
1610       (setq char (decode-char '=ucs code))
1611       (if (encode-char char '=ucs 'defined-only)
1612           (insert-char-data-with-variant char nil 'no-ucs-unified
1613                                          script excluded-script))
1614       (setq code (1+ code)))))
1615
1616 (defun write-char-range-data-to-file (min max file
1617                                           &optional script excluded-script)
1618   (let ((coding-system-for-write char-db-file-coding-system))
1619     (with-temp-buffer
1620       (insert (format ";; -*- coding: %s -*-\n"
1621                       char-db-file-coding-system))
1622       (insert-char-range-data min max script excluded-script)
1623       (write-region (point-min)(point-max) file))))
1624
1625 (defvar what-character-original-window-configuration)
1626
1627 ;;;###autoload
1628 (defun what-char-definition (char)
1629   (interactive (list (char-after)))
1630   (let ((buf (get-buffer-create "*Character Description*"))
1631         (the-buf (current-buffer))
1632         (win-conf (current-window-configuration)))
1633     (pop-to-buffer buf)
1634     (make-local-variable 'what-character-original-window-configuration)
1635     (setq what-character-original-window-configuration win-conf)
1636     (setq buffer-read-only nil)
1637     (erase-buffer)
1638     (condition-case err
1639         (progn
1640           (insert-char-data-with-variant char 'printable)
1641           (unless (char-attribute-alist char)
1642             (insert (format ";; = %c\n"
1643                             (let* ((rest (split-char char))
1644                                    (ccs (pop rest))
1645                                    (code (pop rest)))
1646                               (while rest
1647                                 (setq code (logior (lsh code 8)
1648                                                    (pop rest))))
1649                               (decode-char ccs code)))))
1650           ;; (char-db-update-comment)
1651           (set-buffer-modified-p nil)
1652           (view-mode the-buf (lambda (buf)
1653                                (set-window-configuration
1654                                 what-character-original-window-configuration)
1655                                ))
1656           (goto-char (point-min)))
1657       (error (progn
1658                (set-window-configuration
1659                 what-character-original-window-configuration)
1660                (signal (car err) (cdr err)))))))
1661
1662
1663 ;;; @ end
1664 ;;;
1665
1666 (provide 'char-db-util)
1667
1668 ;;; char-db-util.el ends here