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 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 ?廾 ?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 ?𠧪 ?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     =jef-china3
 346     =shinjigen
 347     =>cns11643-5
 348     =>cns11643-7
 349     =>gt
 350     =>gt-k
 351     =>daikanwa
 352     =>daikanwa/ho
 353     =>cbeta
 354     =ucs-var-001
 355     =ucs-var-002
 356     =ucs-var-003
 357     =ucs-var-004
 358     =ucs-var-005
 359     =ucs-var-006
 360     =ucs-var-007
 361     =ucs-var-008
 362     =ucs-var-010
 363     =ucs-itaiji-001
 364     =ucs-itaiji-002
 365     =ucs-itaiji-003
 366     =ucs-itaiji-004
 367     =ucs-itaiji-005
 368     =ucs-itaiji-006
 369     =ucs-itaiji-007
 370     =ucs-itaiji-008
 371     =ucs-itaiji-009
 372     =ucs-itaiji-010
 373     =ucs-itaiji-011
 374     =ucs-itaiji-012
 375     =+>ucs-var-003
 376     =>ucs-itaiji-001
 377     =>ucs-itaiji-002
 378     =>ucs-itaiji-003
 379     =>ucs-itaiji-004
 380     =>ucs-itaiji-005
 381     =>ucs-itaiji-006
 382     =>ucs-itaiji-007
 383     =>ucs-itaiji-008
 384     =>ucs-itaiji-009
 385     =>ucs-itaiji-010
 386     =>ucs-itaiji-011
 387     =big5-cdp-var-001
 388     =big5-cdp-var-002
 389     =big5-cdp-var-003
 390     =big5-cdp-var-004
 391     =big5-cdp-var-005
 392     =big5-cdp-var-007
 393     =big5-cdp-var-010
 394     =big5-cdp-itaiji-001
 395     =big5-cdp-itaiji-002
 396     =big5-cdp-itaiji-003
 397     =big5-cdp-itaiji-004
 398     =big5-cdp-itaiji-005
 399     =>big5-cdp-itaiji-001
 400     =>big5-cdp-itaiji-002
 401     =>big5-cdp-itaiji-003
 402     =>big5-cdp-itaiji-004
 403     =>big5-cdp-var-001
 404     =>ucs@iwds-1
 405     =>ucs@cognate
 406     =>ucs@component
 407     =>ucs@iwds-1/normalized
 408     =>ucs-itaiji-001@iwds-1
 409     =>ucs-itaiji-002@iwds-1
 410     =>big5-cdp@component
 411     =>big5-cdp@cognate
 412     ==ucs@gb
 413     =ucs@gb
 414     ==ucs-var-002
 415     =ucs@JP/hanazono
 416     =koseki))
 417
 418
 419 ;;; @ char-db formatters
 420 ;;;
 421
 422 (defun char-db-make-char-spec (char)
 423   (let (ret char-spec)
 424     (cond ((characterp char)
 425            (cond ((and (setq ret (encode-char char '=ucs 'defined-only))
 426                        (not (and (<= #xE000 ret)(<= ret #xF8FF))))
 427                   (setq char-spec (list (cons '=ucs ret)))
 428                   (cond ((setq ret (get-char-attribute char 'name))
 429                          (setq char-spec (cons (cons 'name ret) char-spec))
 430                          )
 431                         ((setq ret (get-char-attribute char 'name*))
 432                          (setq char-spec (cons (cons 'name* ret) char-spec))
 433                          ))
 434                   )
 435                  ((encode-char char '=mj 'defined-only)
 436                   (setq char-spec nil)
 437                   (dolist (ccs (charset-list))
 438                     (if (and (or (eq ccs '=mj)
 439                                  ;; (eq (charset-property ccs 'iso-ir) 177)
 440                                  (string-match "=ucs@" (symbol-name ccs))
 441                                  )
 442                              (setq ccs (charset-name ccs))
 443                              (null (assq ccs char-spec))
 444                              (setq ret (encode-char char ccs 'defined-only)))
 445                         (setq char-spec (cons (cons ccs ret) char-spec))))
 446                   )
 447                  ((encode-char char '==mj 'defined-only)
 448                   (setq char-spec nil)
 449                   (dolist (ccs (charset-list))
 450                     (if (and (or (eq ccs '==mj)
 451                                  ;; (eq (charset-property ccs 'iso-ir) 177)
 452                                  (string-match "=ucs@" (symbol-name ccs))
 453                                  )
 454                              (setq ccs (charset-name ccs))
 455                              (null (assq ccs char-spec))
 456                              (setq ret (encode-char char ccs 'defined-only)))
 457                         (setq char-spec (cons (cons ccs ret) char-spec))))
 458                   )
 459                  ((encode-char char '=adobe-japan1 'defined-only)
 460                   (setq char-spec nil)
 461                   (dolist (ccs (charset-list))
 462                     (if (and (or (memq ccs
 463                                        '(=adobe-japan1-0
 464                                          =adobe-japan1-1
 465                                          =adobe-japan1-2
 466                                          =adobe-japan1-3
 467                                          =adobe-japan1-4
 468                                          =adobe-japan1-5
 469                                          =adobe-japan1-6
 470                                          ))
 471                                  ;; (eq (charset-property ccs 'iso-ir) 177)
 472                                  (string-match "=ucs@" (symbol-name ccs))
 473                                  )
 474                              (setq ccs (charset-name ccs))
 475                              (null (assq ccs char-spec))
 476                              (setq ret (encode-char char ccs 'defined-only)))
 477                         (setq char-spec (cons (cons ccs ret) char-spec))))
 478                   )
 479                  ((encode-char char '==adobe-japan1 'defined-only)
 480                   (setq char-spec nil)
 481                   (dolist (ccs (charset-list))
 482                     (if (and (or (memq ccs
 483                                        '(==adobe-japan1-0
 484                                          ==adobe-japan1-1
 485                                          ==adobe-japan1-2
 486                                          ==adobe-japan1-3
 487                                          ==adobe-japan1-4
 488                                          ==adobe-japan1-5
 489                                          ==adobe-japan1-6
 490                                          ))
 491                                  ;; (eq (charset-property ccs 'iso-ir) 177)
 492                                  (string-match "=ucs@" (symbol-name ccs))
 493                                  )
 494                              (setq ccs (charset-name ccs))
 495                              (null (assq ccs char-spec))
 496                              (setq ret (encode-char char ccs 'defined-only)))
 497                         (setq char-spec (cons (cons ccs ret) char-spec))))
 498                   )
 499                  ((setq ret
 500                         (catch 'tag
 501                           (let ((rest char-db-coded-charset-priority-list)
 502                                 ccs)
 503                             (while rest
 504                               (setq ccs (charset-name
 505                                          (find-charset (car rest))))
 506                               (if (setq ret
 507                                         (encode-char char ccs
 508                                                      'defined-only))
 509                                   (throw 'tag (cons ccs ret)))
 510                               (setq rest (cdr rest))))))
 511                   (setq char-spec (list ret))
 512                   (dolist (ccs (delq (car ret) (charset-list)))
 513                     (if (and (or (charset-iso-final-char ccs)
 514                                  (memq ccs
 515                                        '(=daikanwa
 516                                          =daikanwa@rev2
 517                                          ;; =gt-k
 518                                          =jis-x0208@1997
 519                                          ))
 520                                  (and (eq (charset-property ccs 'iso-ir) 177)
 521                                       (not (eq (charset-name ccs) '=ucs@big5)))
 522                                  ;; (string-match "=ucs@" (symbol-name ccs))
 523                                  )
 524                              (setq ccs (charset-name ccs))
 525                              (null (assq ccs char-spec))
 526                              (setq ret (encode-char char ccs 'defined-only)))
 527                         (setq char-spec (cons (cons ccs ret) char-spec))))
 528                   (if (null char-spec)
 529                       (setq char-spec (split-char char)))
 530                   (cond ((setq ret (get-char-attribute char 'name))
 531                          (setq char-spec (cons (cons 'name ret) char-spec))
 532                          )
 533                         ((setq ret (get-char-attribute char 'name*))
 534                          (setq char-spec (cons (cons 'name* ret) char-spec))
 535                          ))
 536                   )
 537                  ((setq ret (get-char-attribute
 538                              char 'ideographic-combination))
 539                   (setq char-spec
 540                         (cons (cons 'ideographic-combination ret)
 541                               char-spec))
 542                   ))
 543            char-spec)
 544           ((consp char)
 545            char))))
 546
 547 (defun char-db-insert-char-spec (char &optional readable column
 548                                       required-features)
 549   (unless column
 550     (setq column (current-column)))
 551   (let (char-spec temp-char)
 552     (setq char-spec (char-db-make-char-spec char))
 553     (unless (or (characterp char) ; char
 554                 (condition-case nil
 555                     (setq char (find-char char-spec))
 556                   (error nil)))
 557       ;; define temporary character
 558       ;;   Current implementation is dirty.
 559       (setq temp-char (define-char (cons '(ideograph-daikanwa . 0)
 560                                          char-spec)))
 561       (remove-char-attribute temp-char 'ideograph-daikanwa)
 562       (setq char temp-char))
 563     (insert-char-attributes char
 564                             readable
 565                             (union (mapcar #'car char-spec)
 566                                    required-features)
 567                             nil 'for-sub-node)
 568     (when temp-char
 569       ;; undefine temporary character
 570       ;;   Current implementation is dirty.
 571       (setq char-spec (char-attribute-alist temp-char))
 572       (while char-spec
 573         (remove-char-attribute temp-char (car (car char-spec)))
 574         (setq char-spec (cdr char-spec))))))
 575
 576 (defun char-db-insert-alist (alist &optional readable column)
 577   (unless column
 578     (setq column (current-column)))
 579   (let ((line-breaking
 580          (concat "\n" (make-string (1+ column) ?\ )))
 581         name value
 582         ret al ; cal
 583         key
 584         lbs cell rest separator)
 585     (insert "(")
 586     (while alist
 587       (setq name (car (car alist))
 588             value (cdr (car alist)))
 589       (cond ((eq name 'char)
 590              (insert "(char . ")
 591              (if (setq ret (condition-case nil
 592                                (find-char value)
 593                              (error nil)))
 594                  (progn
 595                    (setq al nil
 596                          ;; cal nil
 597                          )
 598                    (while value
 599                      (setq key (car (car value)))
 600                      ;; (if (find-charset key)
 601                      ;;     (setq cal (cons key cal))
 602                      (setq al (cons key al))
 603                      ;; )
 604                      (setq value (cdr value)))
 605                    (insert-char-attributes ret
 606                                            readable
 607                                            (or al 'none) ; cal
 608                                            nil 'for-sub-node))
 609                (insert (prin1-to-string value)))
 610              (insert ")")
 611              (insert line-breaking))
 612             ((consp value)
 613              (insert (format "(%-18s " name))
 614              (setq lbs (concat "\n" (make-string (current-column) ?\ )))
 615              (while (consp value)
 616                (setq cell (car value))
 617                (if (and (consp cell)
 618                         (consp (car cell))
 619                         (setq ret (condition-case nil
 620                                       (find-char cell)
 621                                     (error nil)))
 622                         )
 623                    (progn
 624                      (setq rest cell
 625                            al nil
 626                            ;; cal nil
 627                            )
 628                      (while rest
 629                        (setq key (car (car rest)))
 630                        ;; (if (find-charset key)
 631                        ;;     (setq cal (cons key cal))
 632                        (setq al (cons key al))
 633                        ;; )
 634                        (setq rest (cdr rest)))
 635                      (if separator
 636                          (insert lbs))
 637                      (insert-char-attributes ret
 638                                              readable
 639                                              al ; cal
 640                                              nil 'for-sub-node)
 641                      (setq separator lbs))
 642                  (if separator
 643                      (insert separator))
 644                  (insert (prin1-to-string cell))
 645                  (setq separator " "))
 646                (setq value (cdr value)))
 647              (insert ")")
 648              (insert line-breaking))
 649             (t
 650              (insert (format "(%-18s . %S)%s"
 651                              name value
 652                              line-breaking))))
 653       (setq alist (cdr alist))))
 654   (insert ")"))
 655
 656 (defun char-db-insert-char-reference (plist &optional readable column)
 657   (unless column
 658     (setq column (current-column)))
 659   (let ((line-breaking
 660          (concat "\n" (make-string (1+ column) ?\ )))
 661         (separator "")
 662         name value)
 663     (insert "(")
 664     (while plist
 665       (setq name (pop plist))
 666       (setq value (pop plist))
 667       (cond ((eq name :char)
 668              (insert separator)
 669              (insert ":char\t")
 670              (cond ((numberp value)
 671                     (setq value (decode-char '=ucs value)))
 672                    ;; ((consp value)
 673                    ;;  (setq value (or (find-char value)
 674                    ;;                  value)))
 675                    )
 676              (char-db-insert-char-spec value readable)
 677              (insert line-breaking)
 678              (setq separator ""))
 679             ((eq name :radical)
 680              (insert (format "%s%s\t%d ; %c%s"
 681                              separator
 682                              name value
 683                              (ideographic-radical value)
 684                              line-breaking))
 685              (setq separator ""))
 686             (t
 687              (insert (format "%s%s\t%S" separator name value))
 688              (setq separator line-breaking)))
 689       ))
 690   (insert ")"))
 691
 692 (defun char-db-decode-isolated-char (ccs code-point)
 693   (let (ret)
 694     (setq ret
 695           (cond ((eq ccs 'arabic-iso8859-6)
 696                  (decode-char ccs code-point))
 697                 ;; ((eq ccs '=gt)
 698                 ;;  (decode-builtin-char '==gt code-point))
 699                 ((and (memq ccs '(=gt-pj-1
 700                                   =gt-pj-2
 701                                   =gt-pj-3
 702                                   =gt-pj-4
 703                                   =gt-pj-5
 704                                   =gt-pj-6
 705                                   =gt-pj-7
 706                                   =gt-pj-8
 707                                   =gt-pj-9
 708                                   =gt-pj-10
 709                                   =gt-pj-11))
 710                       (setq ret (decode-char ccs code-point))
 711                       (setq ret (encode-char ret '=gt 'defined-only)))
 712                  (decode-builtin-char '=gt ret))
 713                 (t
 714                  (decode-builtin-char ccs code-point))))
 715     (cond ((null ret)
 716            (or (decode-char ccs code-point)
 717                (define-char (list (cons ccs code-point)))))
 718           ((and (<= 0 (char-int ret))
 719                 (<= (char-int ret) #x1F))
 720            (decode-char '=ucs (+ #x2400 (char-int ret))))
 721           ((= (char-int ret) #x7F)
 722            ?\u2421)
 723           (t ret))))
 724
 725 (defvar char-db-convert-obsolete-format t)
 726
 727 (defun char-db-insert-ccs-feature (name value line-breaking)
 728   (cond
 729    ((integerp value)
 730     (insert
 731      (format
 732       (cond
 733        ((memq name '(=>iwds-1
 734                      ==shinjigen
 735                      =shinjigen
 736                      =shinjigen@1ed ==shinjigen@1ed
 737                      =shinjigen@rev ==shinjigen@rev
 738                      =shinjigen/+p@rev ==shinjigen/+p@rev
 739                      ===daikanwa/ho ==daikanwa/ho
 740                      =daikanwa/ho =>>daikanwa/ho =>daikanwa/ho))
 741         "(%-18s .  %04d)\t; %c")
 742        ((eq name '=shinjigen@1ed/24pr)
 743         "(%-18s . %04d)\t; %c")
 744        ((or
 745          (memq name
 746                '(===daikanwa
 747                  ==daikanwa =daikanwa =>>daikanwa =>daikanwa
 748                  =daikanwa@rev1 =daikanwa@rev2
 749                  =daikanwa/+p ==daikanwa/+p ===daikanwa/+p
 750                  =>>daikanwa/+p
 751                  =daikanwa/+2p =>>daikanwa/+2p
 752                  =gt ==gt ===gt
 753                  =>>gt =+>gt =>gt
 754                  =gt-k ==gt-k ===gt-k
 755                  =>>gt-k =>gt-k
 756                  =adobe-japan1-0 ==adobe-japan1-0 ===adobe-japan1-0
 757                  =adobe-japan1-1 ==adobe-japan1-1 ===adobe-japan1-1
 758                  =adobe-japan1-2 ==adobe-japan1-2 ===adobe-japan1-2
 759                  =adobe-japan1-3 ==adobe-japan1-3 ===adobe-japan1-3
 760                  =adobe-japan1-4 ==adobe-japan1-4 ===adobe-japan1-4
 761                  =adobe-japan1-5 ==adobe-japan1-5 ===adobe-japan1-5
 762                  =adobe-japan1-6 ==adobe-japan1-6 ===adobe-japan1-6
 763                  =>>adobe-japan1-0 =+>adobe-japan1-0
 764                  =>>adobe-japan1-1 =+>adobe-japan1-1
 765                  =>>adobe-japan1-2 =+>adobe-japan1-2
 766                  =>>adobe-japan1-3 =+>adobe-japan1-3
 767                  =>>adobe-japan1-4 =+>adobe-japan1-4
 768                  =>>adobe-japan1-5 =+>adobe-japan1-5
 769                  =>>adobe-japan1-6 =+>adobe-japan1-6
 770                  =>cbeta =cbeta =>>cbeta ==cbeta ===cbeta
 771                  =zinbun-oracle =>zinbun-oracle
 772                  =daijiten ==daijiten ===daijiten ==>daijiten
 773                  ===hng-jou ===hng-keg ===hng-dng ===hng-mam
 774                  ===hng-drt ===hng-kgk ===hng-myz ===hng-kda
 775                  ===hng-khi ===hng-khm ===hng-fhs ===hng-hok
 776                  ===hng-kyd ===hng-sok
 777                  ===hng-yhk ===hng-kak ===hng-kar ===hng-kae
 778                  ===hng-sys ===hng-tsu ===hng-tzj
 779                  ===hng-hos ===hng-kkh ===hng-nak ===hng-jhk
 780                  ===hng-hod ===hng-gok ===hng-ink ===hng-nto
 781                  ===hng-nkm ===hng-k24 ===hng-ini ===hng-nkk
 782                  ===hng-kcc ===hng-kcj ===hng-kbk ===hng-sik
 783                  ===hng-skk ===hng-kyu ===hng-ksk ===hng-wan
 784                  ===hng-okd ===hng-wad ===hng-kmi
 785                  ===hng-zkd ===hng-doh ===hng-jyu ===hng-tzs
 786                  ===hng-sai ===hng-kad ===hng-kss
 787                  ===hng-kyo ===hng-ykk ===hng-saa
 788                  ===hng-sab ===hng-wks ===hng-wke ===hng-smk
 789                  ===hng-sgs ===hng-sts
 790                  =shuowen-jiguge ===shuowen-jiguge4 ===shuowen-jiguge5))
 791          ;; (string-match "^=adobe-" (symbol-name name))
 792          )
 793         "(%-18s . %05d)\t; %c")
 794        ((memq name '(=hanyo-denshi/ks
 795                      ==hanyo-denshi/ks ===hanyo-denshi/ks
 796                      =>>hanyo-denshi/ks
 797                      =koseki ==koseki
 798                      =mj ==mj ===mj =>>mj =>mj
 799                      =zihai mojikyo))
 800         "(%-18s . %06d)\t; %c")
 801        ((memq name '(=hanyo-denshi/tk ==hanyo-denshi/tk))
 802         "(%-18s . %08d)\t; %c")
 803        ((>= (charset-dimension name) 2)
 804         "(%-18s . #x%04X)\t; %c")
 805        (t
 806         "(%-18s . #x%02X)\t; %c"))
 807       name
 808       (if (= (charset-iso-graphic-plane name) 1)
 809           (logior value
 810                   (cond ((= (charset-dimension name) 1)
 811                          #x80)
 812                         ((= (charset-dimension name) 2)
 813                          #x8080)
 814                         ((= (charset-dimension name) 3)
 815                          #x808080)
 816                         (t 0)))
 817         value)
 818       (char-db-decode-isolated-char name value)))
 819     (if (and (= (charset-chars name) 94)
 820              (= (charset-dimension name) 2))
 821         (insert (format " [%02d-%02d]"
 822                         (- (lsh value -8) 32)
 823                         (- (logand value 255) 32))))
 824     )
 825    (t
 826     (insert (format "(%-18s . %s)" name value))
 827     ))
 828   (insert line-breaking))
 829
 830 (defun char-db-insert-relation-feature (char name value line-breaking
 831                                              ccss readable)
 832   (insert (format "(%-18s%s " name line-breaking))
 833   (let ((lbs (concat "\n" (make-string (current-column) ?\ )))
 834         separator cell sources required-features
 835         ret)
 836     (while (consp value)
 837       (setq cell (car value))
 838       (if (integerp cell)
 839           (setq cell (decode-char '=ucs cell)))
 840       (cond
 841        ((eq name '->subsumptive)
 842         (when (or (not (some (lambda (atr)
 843                                (get-char-attribute cell atr))
 844                              char-db-ignored-attributes))
 845                   (some (lambda (ccs)
 846                           (encode-char cell ccs 'defined-only))
 847                         ccss))
 848           (if separator
 849               (insert lbs))
 850           (let ((char-db-ignored-attributes
 851                  (cons '<-subsumptive
 852                        char-db-ignored-attributes)))
 853             (insert-char-attributes cell readable nil nil 'for-sub-node))
 854           (setq separator lbs))
 855         )
 856        ((characterp cell)
 857         (setq sources
 858               (get-char-attribute
 859                char (intern (format "%s*sources" name))))
 860         (setq required-features nil)
 861         (dolist (source sources)
 862           (cond
 863            ((memq source '(JP
 864                            JP/Jouyou
 865                            shinjigen shinjigen@1ed shinjigen@rev))
 866             (setq required-features
 867                   (union required-features
 868                          '(=jis-x0208
 869                            =jis-x0208@1990
 870                            =jis-x0213-1@2000
 871                            =jis-x0213-1@2004
 872                            =jis-x0213-2
 873                            =jis-x0212
 874                            =jis-x0208@1983
 875                            =jis-x0208@1978
 876                            =shinjigen
 877                            =shinjigen@1ed
 878                            =shinjigen@rev
 879                            =shinjigen/+p@rev))))
 880            ((eq source 'CN)
 881             (setq required-features
 882                   (union required-features
 883                          '(=gb2312
 884                            =gb12345
 885                            =iso-ir165)))))
 886           (cond
 887            ((find-charset (setq ret (intern (format "=%s" source))))
 888             (setq required-features
 889                   (cons ret required-features)))
 890            (t (setq required-features
 891                     (cons source required-features)))))
 892         (cond ((string-match "@JP" (symbol-name name))
 893                (setq required-features
 894                      (union required-features
 895                             '(=jis-x0208
 896                               =jis-x0208@1990
 897                               =jis-x0213-1-2000
 898                               =jis-x0213-2-2000
 899                               =jis-x0212
 900                               =jis-x0208@1983
 901                               =jis-x0208@1978))))
 902               ((string-match "@CN" (symbol-name name))
 903                (setq required-features
 904                      (union required-features
 905                             '(=gb2312
 906                               =gb12345
 907                               =iso-ir165)))))
 908         (if separator
 909             (insert lbs))
 910         (if readable
 911             (insert (format "%S" cell))
 912           (char-db-insert-char-spec cell readable
 913                                     nil
 914                                     required-features))
 915         (setq separator lbs))
 916        ((consp cell)
 917         (if separator
 918             (insert lbs))
 919         (if (consp (car cell))
 920             (char-db-insert-char-spec cell readable)
 921           (char-db-insert-char-reference cell readable))
 922         (setq separator lbs))
 923        (t
 924         (if separator
 925             (insert separator))
 926         (insert (prin1-to-string cell))
 927         (setq separator " ")))
 928       (setq value (cdr value)))
 929     (insert ")")
 930     (insert line-breaking)))
 931
 932 (defun insert-char-attributes (char &optional readable attributes column
 933                                     for-sub-node)
 934   (unless column
 935     (setq column (current-column)))
 936   (let (name value ; has-long-ccs-name
 937         rest
 938         radical strokes
 939         (line-breaking
 940          (concat "\n" (make-string (1+ column) ?\ )))
 941         lbs cell separator ret
 942         key al cal
 943         dest-ccss ; sources required-features
 944         ccss)
 945     (let (atr-d)
 946       (setq attributes
 947             (sort (if attributes
 948                       (if (consp attributes)
 949                           (progn
 950                             (dolist (name attributes)
 951                               (unless (memq name char-db-ignored-attributes)
 952                                 (if (find-charset name)
 953                                     (push name ccss))
 954                                 (push name atr-d)))
 955                             atr-d))
 956                     (dolist (name (char-attribute-list))
 957                       (unless (memq name char-db-ignored-attributes)
 958                         (if (find-charset name)
 959                             (push name ccss))
 960                         (push name atr-d)))
 961                     atr-d)
 962                   #'char-attribute-name<)))
 963     (insert "(")
 964     (when (memq '<-subsumptive attributes)
 965       (when (or readable (not for-sub-node))
 966         (when (setq value (get-char-attribute char '<-subsumptive))
 967           (char-db-insert-relation-feature char '<-subsumptive value
 968                                            line-breaking
 969                                            ccss readable)))
 970       (setq attributes (delq '<-subsumptive attributes)))
 971     (when (and (memq '<-denotational attributes)
 972                (setq value (get-char-attribute char '<-denotational)))
 973       (char-db-insert-relation-feature char '<-denotational value
 974                                        line-breaking
 975                                        ccss readable)
 976       (setq attributes (delq '<-denotational attributes)))
 977     (when (and (memq '<-denotational@component attributes)
 978                (setq value (get-char-attribute char '<-denotational@component)))
 979       (char-db-insert-relation-feature char '<-denotational@component value
 980                                        line-breaking
 981                                        ccss readable)
 982       (setq attributes (delq '<-denotational@component attributes)))
 983     (when (and (memq '<-denotational@usage attributes)
 984                (setq value (get-char-attribute char '<-denotational@usage)))
 985       (char-db-insert-relation-feature char '<-denotational@usage value
 986                                        line-breaking
 987                                        ccss readable)
 988       (setq attributes (delq '<-denotational@usage attributes)))
 989     (when (and (memq 'name attributes)
 990                (setq value (get-char-attribute char 'name)))
 991       (insert (format
 992                (if (> (+ (current-column) (length value)) 48)
 993                    "(name . %S)%s"
 994                  "(name               . %S)%s")
 995                value line-breaking))
 996       (setq attributes (delq 'name attributes))
 997       )
 998     (when (and (memq 'name* attributes)
 999                (setq value (get-char-attribute char 'name*)))
1000       (insert (format
1001                (if (> (+ (current-column) (length value)) 48)
1002                    "(name* . %S)%s"
1003                  "(name*              . %S)%s")
1004                value line-breaking))
1005       (setq attributes (delq 'name* attributes))
1006       )
1007     (when (and (memq 'script attributes)
1008                (setq value (get-char-attribute char 'script)))
1009       (insert (format "(script\t\t%s)%s"
1010                       (mapconcat (function prin1-to-string)
1011                                  value " ")
1012                       line-breaking))
1013       (setq attributes (delq 'script attributes))
1014       )
1015     (dolist (name '(=>ucs =>ucs*))
1016       (when (and (memq name attributes)
1017                  (setq value (get-char-attribute char name)))
1018         (insert (format "(%-18s . #x%04X)\t; %c%s"
1019                         name value (decode-char '=ucs value)
1020                         line-breaking))
1021         (setq attributes (delq name attributes))))
1022     (dolist (name '(=>ucs@gb =>ucs@big5))
1023       (when (and (memq name attributes)
1024                  (setq value (get-char-attribute char name)))
1025         (insert (format "(%-18s . #x%04X)\t; %c%s"
1026                         name value
1027                         (decode-char (intern
1028                                       (concat "="
1029                                               (substring
1030                                                (symbol-name name) 2)))
1031                                      value)
1032                         line-breaking))
1033         (setq attributes (delq name attributes))
1034         ))
1035     ;; (dolist (name '(=>daikanwa))
1036     ;;   (when (and (memq name attributes)
1037     ;;              (setq value (get-char-attribute char name)))
1038     ;;     (insert
1039     ;;      (if (integerp value)
1040     ;;          (format "(%-18s . %05d)\t; %c%s"
1041     ;;                  name value (decode-char '=daikanwa value)
1042     ;;                  line-breaking)
1043     ;;        (format "(%-18s %s)\t; %c%s"
1044     ;;                name
1045     ;;                (mapconcat (function prin1-to-string)
1046     ;;                           value " ")
1047     ;;                (char-representative-of-daikanwa char)
1048     ;;                line-breaking)))
1049     ;;     (setq attributes (delq name attributes))))
1050     (when (and (memq 'general-category attributes)
1051                (setq value (get-char-attribute char 'general-category)))
1052       (insert (format
1053                "(general-category\t%s) ; %s%s"
1054                (mapconcat (lambda (cell)
1055                             (format "%S" cell))
1056                           value " ")
1057                (cond ((rassoc value unidata-normative-category-alist)
1058                       "Normative Category")
1059                      ((rassoc value unidata-informative-category-alist)
1060                       "Informative Category")
1061                      (t
1062                       "Unknown Category"))
1063                line-breaking))
1064       (setq attributes (delq 'general-category attributes))
1065       )
1066     (when (and (memq 'bidi-category attributes)
1067                (setq value (get-char-attribute char 'bidi-category)))
1068       (insert (format "(bidi-category\t. %S)%s"
1069                       value
1070                       line-breaking))
1071       (setq attributes (delq 'bidi-category attributes))
1072       )
1073     (unless (or (not (memq 'mirrored attributes))
1074                 (eq (setq value (get-char-attribute char 'mirrored 'empty))
1075                     'empty))
1076       (insert (format "(mirrored\t\t. %S)%s"
1077                       value
1078                       line-breaking))
1079       (setq attributes (delq 'mirrored attributes))
1080       )
1081     (cond
1082      ((and (memq 'decimal-digit-value attributes)
1083            (setq value (get-char-attribute char 'decimal-digit-value)))
1084       (insert (format "(decimal-digit-value . %S)%s"
1085                       value
1086                       line-breaking))
1087       (setq attributes (delq 'decimal-digit-value attributes))
1088       (when (and (memq 'digit-value attributes)
1089                  (setq value (get-char-attribute char 'digit-value)))
1090         (insert (format "(digit-value\t . %S)%s"
1091                         value
1092                         line-breaking))
1093         (setq attributes (delq 'digit-value attributes))
1094         )
1095       (when (and (memq 'numeric-value attributes)
1096                  (setq value (get-char-attribute char 'numeric-value)))
1097         (insert (format "(numeric-value\t . %S)%s"
1098                         value
1099                         line-breaking))
1100         (setq attributes (delq 'numeric-value attributes))
1101         )
1102       )
1103      (t
1104       (when (and (memq 'digit-value attributes)
1105                  (setq value (get-char-attribute char 'digit-value)))
1106         (insert (format "(digit-value\t. %S)%s"
1107                         value
1108                         line-breaking))
1109         (setq attributes (delq 'digit-value attributes))
1110         )
1111       (when (and (memq 'numeric-value attributes)
1112                  (setq value (get-char-attribute char 'numeric-value)))
1113         (insert (format "(numeric-value\t. %S)%s"
1114                         value
1115                         line-breaking))
1116         (setq attributes (delq 'numeric-value attributes))
1117         )))
1118     (when (and (memq 'iso-10646-comment attributes)
1119                (setq value (get-char-attribute char 'iso-10646-comment)))
1120       (insert (format "(iso-10646-comment\t. %S)%s"
1121                       value
1122                       line-breaking))
1123       (setq attributes (delq 'iso-10646-comment attributes))
1124       )
1125     (when (and (memq 'morohashi-daikanwa attributes)
1126                (setq value (get-char-attribute char 'morohashi-daikanwa)))
1127       (insert (format "(morohashi-daikanwa\t%s)%s"
1128                       (mapconcat (function prin1-to-string) value " ")
1129                       line-breaking))
1130       (setq attributes (delq 'morohashi-daikanwa attributes))
1131       )
1132     (setq radical nil
1133           strokes nil)
1134     (when (and (memq 'ideographic-radical attributes)
1135                (setq value (get-char-attribute char 'ideographic-radical)))
1136       (setq radical value)
1137       (insert (format "(ideographic-radical . %S)\t; %c%s"
1138                       radical
1139                       (ideographic-radical radical)
1140                       line-breaking))
1141       (setq attributes (delq 'ideographic-radical attributes))
1142       )
1143     (when (and (memq 'shuowen-radical attributes)
1144                (setq value (get-char-attribute char 'shuowen-radical)))
1145       (insert (format "(shuowen-radical\t. %S)\t; %c%s"
1146                       value
1147                       (shuowen-radical value)
1148                       line-breaking))
1149       (setq attributes (delq 'shuowen-radical attributes))
1150       )
1151     (let (key)
1152       (dolist (domain
1153                (append
1154                 char-db-feature-domains
1155                 (let (dest domain)
1156                   (dolist (feature (char-attribute-list))
1157                     (setq feature (symbol-name feature))
1158                     (when (string-match
1159                            "\\(radical\\|strokes\\)@\\([^@*]+\\)\\(\\*\\|$\\)"
1160                            feature)
1161                       (setq domain (intern (match-string 2 feature)))
1162                      (unless (memq domain dest)
1163                        (setq dest (cons domain dest)))))
1164                   (sort dest #'string<))))
1165         (setq key (intern (format "%s@%s" 'ideographic-radical domain)))
1166         (when (and (memq key attributes)
1167                    (setq value (get-char-attribute char key)))
1168           (setq radical value)
1169           (insert (format "(%s . %S)\t; %c%s"
1170                           key
1171                           radical
1172                           (ideographic-radical radical)
1173                           line-breaking))
1174           (setq attributes (delq key attributes))
1175           )
1176         (setq key (intern (format "%s@%s" 'ideographic-strokes domain)))
1177         (when (and (memq key attributes)
1178                    (setq value (get-char-attribute char key)))
1179           (setq strokes value)
1180           (insert (format "(%s . %S)%s"
1181                           key
1182                           strokes
1183                           line-breaking))
1184           (setq attributes (delq key attributes))
1185           )
1186         (setq key (intern (format "%s@%s" 'total-strokes domain)))
1187         (when (and (memq key attributes)
1188                    (setq value (get-char-attribute char key)))
1189           (insert (format "(%s       . %S)%s"
1190                           key
1191                           value
1192                           line-breaking))
1193           (setq attributes (delq key attributes))
1194           )
1195         (dolist (feature '(ideographic-radical
1196                            ideographic-strokes
1197                            total-strokes))
1198           (setq key (intern (format "%s@%s*sources" feature domain)))
1199           (when (and (memq key attributes)
1200                      (setq value (get-char-attribute char key)))
1201             (insert (format "(%s%s" key line-breaking))
1202             (dolist (cell value)
1203               (insert (format " %s" cell)))
1204             (insert ")")
1205             (insert line-breaking)
1206             (setq attributes (delq key attributes))
1207             ))
1208         ))
1209     (when (and (memq 'ideographic-strokes attributes)
1210                (setq value (get-char-attribute char 'ideographic-strokes)))
1211       (setq strokes value)
1212       (insert (format "(ideographic-strokes . %S)%s"
1213                       strokes
1214                       line-breaking))
1215       (setq attributes (delq 'ideographic-strokes attributes))
1216       )
1217     (when (and (memq 'kangxi-radical attributes)
1218                (setq value (get-char-attribute char 'kangxi-radical)))
1219       (unless (eq value radical)
1220         (insert (format "(kangxi-radical\t . %S)\t; %c%s"
1221                         value
1222                         (ideographic-radical value)
1223                         line-breaking))
1224         (or radical
1225             (setq radical value)))
1226       (setq attributes (delq 'kangxi-radical attributes))
1227       )
1228     (when (and (memq 'kangxi-strokes attributes)
1229                (setq value (get-char-attribute char 'kangxi-strokes)))
1230       (unless (eq value strokes)
1231         (insert (format "(kangxi-strokes\t . %S)%s"
1232                         value
1233                         line-breaking))
1234         (or strokes
1235             (setq strokes value)))
1236       (setq attributes (delq 'kangxi-strokes attributes))
1237       )
1238     ;; (when (and (memq 'japanese-radical attributes)
1239     ;;            (setq value (get-char-attribute char 'japanese-radical)))
1240     ;;   (unless (eq value radical)
1241     ;;     (insert (format "(japanese-radical\t . %S)\t; %c%s"
1242     ;;                     value
1243     ;;                     (ideographic-radical value)
1244     ;;                     line-breaking))
1245     ;;     (or radical
1246     ;;         (setq radical value)))
1247     ;;   (setq attributes (delq 'japanese-radical attributes))
1248     ;;   )
1249     (when (and (memq 'japanese-strokes attributes)
1250                (setq value (get-char-attribute char 'japanese-strokes)))
1251       (unless (eq value strokes)
1252         (insert (format "(japanese-strokes\t . %S)%s"
1253                         value
1254                         line-breaking))
1255         (or strokes
1256             (setq strokes value)))
1257       (setq attributes (delq 'japanese-strokes attributes))
1258       )
1259     (when (and (memq 'cns-radical attributes)
1260                (setq value (get-char-attribute char 'cns-radical)))
1261       (insert (format "(cns-radical\t . %S)\t; %c%s"
1262                       value
1263                       (ideographic-radical value)
1264                       line-breaking))
1265       (setq attributes (delq 'cns-radical attributes))
1266       )
1267     (when (and (memq 'cns-strokes attributes)
1268                (setq value (get-char-attribute char 'cns-strokes)))
1269       (unless (eq value strokes)
1270         (insert (format "(cns-strokes\t . %S)%s"
1271                         value
1272                         line-breaking))
1273         (or strokes
1274             (setq strokes value)))
1275       (setq attributes (delq 'cns-strokes attributes))
1276       )
1277     ;; (when (and (memq 'ideographic- attributes)
1278     ;;            (setq value (get-char-attribute char 'ideographic-)))
1279     ;;   (insert "(ideographic-       ")
1280     ;;   (setq lbs (concat "\n" (make-string (current-column) ?\ ))
1281     ;;         separator nil)
1282     ;;   (while (consp value)
1283     ;;     (setq cell (car value))
1284     ;;     (if (integerp cell)
1285     ;;         (setq cell (decode-char '=ucs cell)))
1286     ;;     (cond ((characterp cell)
1287     ;;            (if separator
1288     ;;                (insert lbs))
1289     ;;            (if readable
1290     ;;                (insert (format "%S" cell))
1291     ;;              (char-db-insert-char-spec cell readable))
1292     ;;            (setq separator lbs))
1293     ;;           ((consp cell)
1294     ;;            (if separator
1295     ;;                (insert lbs))
1296     ;;            (if (consp (car cell))
1297     ;;                (char-db-insert-char-spec cell readable)
1298     ;;              (char-db-insert-char-reference cell readable))
1299     ;;            (setq separator lbs))
1300     ;;           (t
1301     ;;            (if separator
1302     ;;                (insert separator))
1303     ;;            (insert (prin1-to-string cell))
1304     ;;            (setq separator " ")))
1305     ;;     (setq value (cdr value)))
1306     ;;   (insert ")")
1307     ;;   (insert line-breaking)
1308     ;;   (setq attributes (delq 'ideographic- attributes)))
1309     (when (and (memq 'total-strokes attributes)
1310                (setq value (get-char-attribute char 'total-strokes)))
1311       (insert (format "(total-strokes       . %S)%s"
1312                       value
1313                       line-breaking))
1314       (setq attributes (delq 'total-strokes attributes))
1315       )
1316     ;; (when (and (memq '->ideograph attributes)
1317     ;;            (setq value (get-char-attribute char '->ideograph)))
1318     ;;   (insert (format "(->ideograph\t%s)%s"
1319     ;;                   (mapconcat (lambda (code)
1320     ;;                                (cond ((symbolp code)
1321     ;;                                       (symbol-name code))
1322     ;;                                      ((integerp code)
1323     ;;                                       (format "#x%04X" code))
1324     ;;                                      (t
1325     ;;                                       (format "%s %S"
1326     ;;                                               line-breaking code))))
1327     ;;                              value " ")
1328     ;;                   line-breaking))
1329     ;;   (setq attributes (delq '->ideograph attributes))
1330     ;;   )
1331     ;; (when (and (memq '->decomposition attributes)
1332     ;;            (setq value (get-char-attribute char '->decomposition)))
1333     ;;   (insert (format "(->decomposition\t%s)%s"
1334     ;;                   (mapconcat (lambda (code)
1335     ;;                                (cond ((symbolp code)
1336     ;;                                       (symbol-name code))
1337     ;;                                      ((characterp code)
1338     ;;                                       (if readable
1339     ;;                                           (format "%S" code)
1340     ;;                                         (format "#x%04X"
1341     ;;                                                 (char-int code))
1342     ;;                                         ))
1343     ;;                                      ((integerp code)
1344     ;;                                       (format "#x%04X" code))
1345     ;;                                      (t
1346     ;;                                       (format "%s%S" line-breaking code))))
1347     ;;                              value " ")
1348     ;;                   line-breaking))
1349     ;;   (setq attributes (delq '->decomposition attributes))
1350     ;;   )
1351     (if (equal (get-char-attribute char '->titlecase)
1352                (get-char-attribute char '->uppercase))
1353         (setq attributes (delq '->titlecase attributes)))
1354     (when (and (memq '->mojikyo attributes)
1355                (setq value (get-char-attribute char '->mojikyo)))
1356       (insert (format "(->mojikyo\t\t. %06d)\t; %c%s"
1357                       value (decode-char 'mojikyo value)
1358                       line-breaking))
1359       (setq attributes (delq '->mojikyo attributes))
1360       )
1361     (unless readable
1362       (dolist (ignored '(composition
1363                          ->denotational <-subsumptive ->ucs-unified
1364                          ;; ->ideographic-component-forms
1365                          ))
1366         (setq attributes (delq ignored attributes))))
1367     (while attributes
1368       (setq name (car attributes))
1369       (unless (eq (setq value (get-char-attribute char name 'value-is-empty))
1370                   'value-is-empty)
1371         (cond ((setq ret (find-charset name))
1372                (setq name (charset-name ret))
1373                (when (not (memq name dest-ccss))
1374                  (setq dest-ccss (cons name dest-ccss))
1375                  (char-db-insert-ccs-feature name value line-breaking))
1376                )
1377               ((string-match "^=>ucs@" (symbol-name name))
1378                (insert (format "(%-18s . #x%04X)\t; %c%s"
1379                                name value (decode-char '=ucs value)
1380                                line-breaking))
1381                )
1382               ((eq name 'jisx0208-1978/4X)
1383                (insert (format "(%-18s . #x%04X)%s"
1384                                name value
1385                                line-breaking))
1386                )
1387               ((and
1388                 (not readable)
1389                 (not (eq name '->subsumptive))
1390                 (not (eq name '->uppercase))
1391                 (not (eq name '->lowercase))
1392                 (not (eq name '->titlecase))
1393                 (not (eq name '->canonical))
1394                 (not (eq name '->Bopomofo))
1395                 (not (eq name '->mistakable))
1396                 (not (eq name '->ideographic-variants))
1397                 (null (get-char-attribute
1398                        char (intern (format "%s*sources" name))))
1399                 (not (string-match "\\*sources$" (symbol-name name)))
1400                 (null (get-char-attribute
1401                        char (intern (format "%s*note" name))))
1402                 (not (string-match "\\*note$" (symbol-name name)))
1403                 (or (eq name '<-identical)
1404                     (eq name '<-uppercase)
1405                     (eq name '<-lowercase)
1406                     (eq name '<-titlecase)
1407                     (eq name '<-canonical)
1408                     (eq name '<-ideographic-variants)
1409                     ;; (eq name '<-synonyms)
1410                     (string-match "^<-synonyms" (symbol-name name))
1411                     (eq name '<-mistakable)
1412                     (when (string-match "^->" (symbol-name name))
1413                       (cond
1414                        ((string-match "^->fullwidth" (symbol-name name))
1415                         (not (and (consp value)
1416                                   (characterp (car value))
1417                                   (encode-char
1418                                    (car value) '=ucs 'defined-only)))
1419                         )
1420                        (t)))
1421                     ))
1422                )
1423               ((or (eq name 'ideographic-structure)
1424                    (eq name 'ideographic-structure@apparent)
1425                    (eq name 'ideographic-combination)
1426                    ;; (eq name 'ideographic-)
1427                    (eq name '=decomposition)
1428                    (char-feature-base-name= '=decomposition name)
1429                    (char-feature-base-name= '=>decomposition name)
1430                    ;; (string-match "^=>*decomposition\\(@[^*]+\\)?$"
1431                    ;;               (symbol-name name))
1432                    (string-match "^\\(->\\|<-\\)[^*]*$" (symbol-name name))
1433                    (string-match "^\\(->\\|<-\\)[^*]*\\*sources$"
1434                                  (symbol-name name))
1435                    )
1436                (char-db-insert-relation-feature char name value
1437                                                 line-breaking
1438                                                 ccss readable))
1439               ;; ((memq name '(ideograph=
1440               ;;               original-ideograph-of
1441               ;;               ancient-ideograph-of
1442               ;;               vulgar-ideograph-of
1443               ;;               wrong-ideograph-of
1444               ;;               ;; simplified-ideograph-of
1445               ;;               ideographic-variants
1446               ;;               ;; ideographic-different-form-of
1447               ;;               ))
1448               ;;  (insert (format "(%-18s%s " name line-breaking))
1449               ;;  (setq lbs (concat "\n" (make-string (current-column) ?\ ))
1450               ;;        separator nil)
1451               ;;  (while (consp value)
1452               ;;    (setq cell (car value))
1453               ;;    (if (and (consp cell)
1454               ;;             (consp (car cell)))
1455               ;;        (progn
1456               ;;          (if separator
1457               ;;              (insert lbs))
1458               ;;          (char-db-insert-alist cell readable)
1459               ;;          (setq separator lbs))
1460               ;;      (if separator
1461               ;;          (insert separator))
1462               ;;      (insert (prin1-to-string cell))
1463               ;;      (setq separator " "))
1464               ;;    (setq value (cdr value)))
1465               ;;  (insert ")")
1466               ;;  (insert line-breaking))
1467               ((consp value)
1468                (insert (format "(%-18s " name))
1469                (setq lbs (concat "\n" (make-string (current-column) ?\ ))
1470                      separator nil)
1471                (while (consp value)
1472                  (setq cell (car value))
1473                  (if (and (consp cell)
1474                           (consp (car cell))
1475                           (setq ret (condition-case nil
1476                                         (find-char cell)
1477                                       (error nil))))
1478                      (progn
1479                        (setq rest cell
1480                              al nil
1481                              cal nil)
1482                        (while rest
1483                          (setq key (car (car rest)))
1484                          (if (find-charset key)
1485                              (setq cal (cons key cal))
1486                            (setq al (cons key al)))
1487                          (setq rest (cdr rest)))
1488                        (if separator
1489                            (insert lbs))
1490                        (insert-char-attributes ret
1491                                                readable
1492                                                al ; cal
1493                                                nil 'for-sub-node)
1494                        (setq separator lbs))
1495                    (setq ret (prin1-to-string cell))
1496                    (if separator
1497                        (if (< (+ (current-column)
1498                                  (length ret)
1499                                  (length separator))
1500                               76)
1501                            (insert separator)
1502                          (insert lbs)))
1503                    (insert ret)
1504                    (setq separator " "))
1505                  (setq value (cdr value)))
1506                (insert ")")
1507                (insert line-breaking))
1508               (t
1509                (insert (format "(%-18s" name))
1510                (setq ret (prin1-to-string value))
1511                (unless (< (+ (current-column)
1512                              (length ret)
1513                              3)
1514                           76)
1515                  (insert line-breaking))
1516                (insert " . " ret ")" line-breaking)
1517                ;; (insert (format "(%-18s . %S)%s"
1518                ;;                 name value
1519                ;;                 line-breaking))
1520                )
1521               ))
1522       (setq attributes (cdr attributes)))
1523     (insert ")")))
1524
1525 (defun insert-char-data (char &optional readable
1526                               attributes)
1527   (save-restriction
1528     (narrow-to-region (point)(point))
1529     (insert "(define-char
1530   '")
1531     (insert-char-attributes char readable attributes)
1532     (insert ")\n")
1533     (goto-char (point-min))
1534     (while (re-search-forward "[ \t]+$" nil t)
1535       (replace-match ""))
1536     ;; from tabify.
1537     (goto-char (point-min))
1538     (while (re-search-forward "[ \t][ \t][ \t]*" nil t)
1539       (let ((column (current-column))
1540             (indent-tabs-mode t))
1541         (delete-region (match-beginning 0) (point))
1542         (indent-to column)))
1543     (goto-char (point-max))
1544     ;; (tabify (point-min)(point-max))
1545     ))
1546
1547 (defun insert-char-data-with-variant (char &optional printable
1548                                            no-ucs-unified
1549                                            script excluded-script)
1550   (insert-char-data char printable)
1551   (let ((variants (char-variants char))
1552         rest
1553         variant vs ret)
1554     (setq variants (sort variants #'<))
1555     (setq rest variants)
1556     (setq variants (cons char variants))
1557     (while rest
1558       (setq variant (car rest))
1559       (unless (get-char-attribute variant '<-subsumptive)
1560         (if (and (or (null script)
1561                      (null (setq vs (get-char-attribute variant 'script)))
1562                      (memq script vs))
1563                  (or (null excluded-script)
1564                      (null (setq vs (get-char-attribute variant 'script)))
1565                      (not (memq excluded-script vs))))
1566             (unless (and no-ucs-unified (get-char-attribute variant '=ucs))
1567               (insert-char-data variant printable)
1568               (if (setq ret (char-variants variant))
1569                   (while ret
1570                     (or (memq (car ret) variants)
1571                         ;; (get-char-attribute (car ret) '<-subsumptive)
1572                         (setq rest (nconc rest (list (car ret)))))
1573                     (setq ret (cdr ret)))))))
1574       (setq rest (cdr rest)))))
1575
1576 (defun insert-char-range-data (min max &optional script excluded-script)
1577   (let ((code min)
1578         char)
1579     (while (<= code max)
1580       (setq char (decode-char '=ucs code))
1581       (if (encode-char char '=ucs 'defined-only)
1582           (insert-char-data-with-variant char nil 'no-ucs-unified
1583                                          script excluded-script))
1584       (setq code (1+ code)))))
1585
1586 (defun write-char-range-data-to-file (min max file
1587                                           &optional script excluded-script)
1588   (let ((coding-system-for-write char-db-file-coding-system))
1589     (with-temp-buffer
1590       (insert (format ";; -*- coding: %s -*-\n"
1591                       char-db-file-coding-system))
1592       (insert-char-range-data min max script excluded-script)
1593       (write-region (point-min)(point-max) file))))
1594
1595 (defvar what-character-original-window-configuration)
1596
1597 ;;;###autoload
1598 (defun what-char-definition (char)
1599   (interactive (list (char-after)))
1600   (let ((buf (get-buffer-create "*Character Description*"))
1601         (the-buf (current-buffer))
1602         (win-conf (current-window-configuration)))
1603     (pop-to-buffer buf)
1604     (make-local-variable 'what-character-original-window-configuration)
1605     (setq what-character-original-window-configuration win-conf)
1606     (setq buffer-read-only nil)
1607     (erase-buffer)
1608     (condition-case err
1609         (progn
1610           (insert-char-data-with-variant char 'printable)
1611           (unless (char-attribute-alist char)
1612             (insert (format ";; = %c\n"
1613                             (let* ((rest (split-char char))
1614                                    (ccs (pop rest))
1615                                    (code (pop rest)))
1616                               (while rest
1617                                 (setq code (logior (lsh code 8)
1618                                                    (pop rest))))
1619                               (decode-char ccs code)))))
1620           ;; (char-db-update-comment)
1621           (set-buffer-modified-p nil)
1622           (view-mode the-buf (lambda (buf)
1623                                (set-window-configuration
1624                                 what-character-original-window-configuration)
1625                                ))
1626           (goto-char (point-min)))
1627       (error (progn
1628                (set-window-configuration
1629                 what-character-original-window-configuration)
1630                (signal (car err) (cdr err)))))))
1631
1632
1633 ;;; @ end
1634 ;;;
1635
1636 (provide 'char-db-util)
1637
1638 ;;; char-db-util.el ends here