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 MORIOKA Tomohiko.
   6
   7 ;; Author: MORIOKA Tomohiko <tomo@kanji.zinbun.kyoto-u.ac.jp>
   8 ;; Keywords: CHISE, Character Database, ISO/IEC 10646, UCS, Unicode, MULE.
   9
  10 ;; This file is part of XEmacs CHISE.
  11
  12 ;; XEmacs CHISE is free software; you can redistribute it and/or
  13 ;; modify it under the terms of the GNU General Public License as
  14 ;; published by the Free Software Foundation; either version 2, or (at
  15 ;; your option) any later version.
  16
  17 ;; XEmacs CHISE is distributed in the hope that it will be useful,
  18 ;; but WITHOUT ANY WARRANTY; without even the implied warranty of
  19 ;; MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the GNU
  20 ;; General Public License for more details.
  21
  22 ;; You should have received a copy of the GNU General Public License
  23 ;; along with XEmacs CHISE; see the file COPYING.  If not, write to
  24 ;; the Free Software Foundation, Inc., 59 Temple Place - Suite 330,
  25 ;; Boston, MA 02111-1307, USA.
  26
  27 ;;; Code:
  28
  29 (require 'chise-subr)
  30 (require 'ideograph-subr)
  31
  32 (defconst unidata-normative-category-alist
  33   '(("Lu" letter        uppercase)
  34     ("Ll" letter        lowercase)
  35     ("Lt" letter        titlecase)
  36     ("Mn" mark          non-spacing)
  37     ("Mc" mark          spacing-combining)
  38     ("Me" mark          enclosing)
  39     ("Nd" number        decimal-digit)
  40     ("Nl" number        letter)
  41     ("No" number        other)
  42     ("Zs" separator     space)
  43     ("Zl" separator     line)
  44     ("Zp" separator     paragraph)
  45     ("Cc" other         control)
  46     ("Cf" other         format)
  47     ("Cs" other         surrogate)
  48     ("Co" other         private-use)
  49     ("Cn" other         not-assigned)))
  50
  51 (defconst unidata-informative-category-alist
  52   '(("Lm" letter        modifier)
  53     ("Lo" letter        other)
  54     ("Pc" punctuation   connector)
  55     ("Pd" punctuation   dash)
  56     ("Ps" punctuation   open)
  57     ("Pe" punctuation   close)
  58     ("Pi" punctuation   initial-quote)
  59     ("Pf" punctuation   final-quote)
  60     ("Po" punctuation   other)
  61     ("Sm" symbol        math)
  62     ("Sc" symbol        currency)
  63     ("Sk" symbol        modifier)
  64     ("So" symbol        other)
  65     ))
  66
  67 (defconst shuowen-radicals
  68   [?一 ?上 ?示 ?三 ?王 ?玉 ?玨 ?气 ?士 ?丨 ; 010
  69    ?屮 ?艸 ?蓐 ?茻 ?小 ?八 ?釆 ?半 ?牛 ?犛 ; 020
  70    ?告 ?口 ?凵 ?吅 ?哭 ?走 ?止 ?癶 ?步 ?此 ; 030
  71    ?正 ?是 ?辵 ?彳 ?廴 ?㢟 ?行 ?齒 ?牙 ?足 ; 040
  72    ?疋 ?品 ?龠 ?冊 ?㗊 ?舌 ?干 ?𧮫 ?只 ?㕯 ; 050
  73    ?句 ?丩 ?古 ?十 ?卅 ?言 ?誩 ?音 ?䇂 ?丵 ; 060
  74    ?菐 ?𠬞 ?𠬜 ?共 ?異 ?舁 ?𦥑 ?䢅 ?爨 ?革 ; 070
  75    ?鬲 ?䰜 ?爪 ?𠃨 ?鬥 ?又 ?𠂇 ?史 ?支 ?𦘒 ; 080
  76    ?聿 ?畫 ?隶 ?臤 ?臣 ?殳 ?殺 ?𠘧 ?寸 ?皮 ; 090
  77    ?㼱 ?攴 ?敎 ?卜 ?用 ?爻 ?㸚 ?𥄎 ?目 ?䀠 ; 100
  78    ?眉 ?盾 ?自 ?白 ?鼻 ?皕 ?習 ?羽 ?隹 ?奞 ; 110
  79    ?雈 ?𦫳 ?𥄕 ?羊 ?羴 ?瞿 ?雔 ?雥 ?鳥 ?烏 ; 120
  80    ?𠦒 ?冓 ?幺 ?𢆶 ?叀 ?玄 ?予 ?放 ?𠬪 ?𣦼 ; 130
  81    ?歺 ?死 ?冎 ?骨 ?肉 ?筋 ?刀 ?刃 ?㓞 ?丯 ; 140
  82    ?耒 ?角 ?竹 ?箕 ?丌 ?左 ?工 ?㠭 ?巫 ?甘 ; 150
  83    ?曰 ?乃 ?丂 ?可 ?兮 ?号 ?亏 ?旨 ?喜 ?壴 ; 160
  84 ;  ?旨 ?曰 ?乃 ?丂 ?可 ?兮 ?号 ?亏 ?喜 ?壴 ; 160
  85    ?鼓 ?豈 ?豆 ?豊 ?豐 ?䖒 ?虍 ?虎 ?虤 ?皿 ; 170
  86    ?𠙴 ?去 ?血 ?丶 ?丹 ?青 ?井 ?皀 ?鬯 ?食 ; 180
  87    ?亼 ?會 ?倉 ?入 ?缶 ?矢 ?高 ?冂 ?𩫏 ?京 ; 190
  88    ?亯 ?𣆪 ?畗 ?㐭 ?嗇 ?來 ?麥 ?夊 ?舛 ?䑞 ; 200
  89    ?韋 ?弟 ?夂 ?久 ?桀 ?木 ?東 ?林 ?才 ?叒 ; 210
  90    ?之 ?帀 ?出 ?𣎵 ?生 ?乇 ?𠂹 ?𠌶 ?華 ?𥝌 ; 220
  91    ?稽 ?巢 ?桼 ?束 ?㯻 ?囗 ?員 ?貝 ?邑 ?𨛜 ; 230
  92    ?日 ?旦 ?倝 ?㫃 ?冥 ?晶 ?月 ?有 ?明 ?囧 ; 240
  93    ?夕 ?多 ?毌 ?𢎘 ?𣐺 ?卣 ?齊 ?朿 ?片 ?鼎 ; 250
  94    ?克 ?彔 ?禾 ?秝 ?黍 ?香 ?米 ?毇 ?臼 ?凶 ; 260
  95    ?𣎳 ?𣏟 ?麻 ?尗 ?耑 ?韭 ?瓜 ?瓠 ?宀 ?宮 ; 270
  96    ?呂 ?穴 ?㝱 ?疒 ?冖 ?𠔼 ?冃 ?㒳 ?网 ?襾 ; 280
  97    ?巾 ?巿 ?帛 ?白 ?㡀 ?黹 ?人 ?𠤎 ?匕 ?从 ; 290
  98    ?比 ?北 ?丘 ?㐺 ?𡈼 ?重 ?臥 ?身 ?㐆 ?衣 ; 300
  99    ?裘 ?老 ?毛 ?毳 ?尸 ?尺 ?尾 ?履 ?舟 ?方 ; 310
 100    ?儿 ?兄 ?兂 ?皃 ?𠑹 ?先 ?秃 ?見 ?覞 ?欠 ; 320
 101    ?㱃 ?㳄 ?旡 ?頁 ?𦣻 ?面 ?丏 ?首 ?𥄉 ?須 ; 330
 102    ?彡 ?彣 ?文 ?髟 ?后 ?司 ?卮 ?卩 ?印 ?色 ; 340
 103    ?𠨍 ?辟 ?勹 ?包 ?茍 ?鬼 ?甶 ?厶 ?嵬 ?山 ; 350
 104    ?屾 ?屵 ?广 ?厂 ?丸 ?危 ?石 ?長 ?勿 ?冄 ; 360
 105    ?而 ?豕 ?㣇 ?彑 ?豚 ?豸 ?𤉡 ?易 ?象 ?馬 ; 370
 106    ?𢊁 ?鹿 ?麤 ?㲋 ?兔 ?萈 ?犬 ?㹜 ?鼠 ?能 ; 380
 107    ?熊 ?火 ?炎 ?黑 ?囪 ?焱 ?炙 ?赤 ?大 ?亦 ; 390
 108    ?夨 ?夭 ?交 ?尣 ?壺 ?壹 ?㚔 ?奢 ?亢 ?夲 ; 400
 109    ?夰 ?亣 ?夫 ?立 ?竝 ?囟 ?思 ?心 ?惢 ?水 ; 410
 110    ?沝 ?瀕 ?𡿨 ?巜 ?川 ?泉 ?灥 ?永 ?𠂢 ?谷 ; 420
 111    ?仌 ?雨 ?雲 ?魚 ?𩺰 ?燕 ?龍 ?飛 ?非 ?卂 ; 430
 112    ?𠃉 ?不 ?至 ?西 ?鹵 ?鹽 ?戶 ?門 ?耳 ?𦣝 ; 440
 113    ?手 ?𠦬 ?女 ?毋 ?民 ?丿 ?𠂆 ?乁 ?氏 ?氐 ; 450
 114    ?戈 ?戉 ?我 ?亅 ?琴 ?𠃊 ?亡 ?匸 ?匚 ?曲 ; 460
 115    ?甾 ?瓦 ?弓 ?弜 ?弦 ?系 ?糸 ?素 ?絲 ?率 ; 470
 116    ?虫 ?䖵 ?蟲 ?風 ?它 ?龜 ?黽 ?卵 ?二 ?土 ; 480
 117    ?垚 ?堇 ?里 ?田 ?畕 ?黃 ?男 ?力 ?劦 ?金 ; 490
 118    ?幵 ?勺 ?几 ?且 ?斤 ?斗 ?矛 ?車 ?𠂤 ?𨸏 ; 500
 119    ?𨺅 ?厽 ?四 ?宁 ?叕 ?亞 ?五 ?六 ?七 ?九 ; 510
 120    ?禸 ?嘼 ?甲 ?乙 ?丙 ?丁 ?戊 ?己 ?巴 ?庚 ; 520
 121    ?辛 ?辡 ?壬 ?癸 ?子 ?了 ?孨 ?𠫓 ?丑 ?寅 ; 530
 122    ?卯 ?辰 ?巳 ?午 ?未 ?申 ?酉 ?酋 ?戌 ?亥 ; 540
 123    ])
 124
 125 (defun shuowen-radical (number)
 126   (aref shuowen-radicals (1- number)))
 127
 128 (defvar char-db-file-coding-system 'utf-8-mcs-er)
 129
 130 (defvar char-db-ignored-attributes '(ideographic-products))
 131
 132 (defvar char-db-coded-charset-priority-list
 133   '(ascii
 134     control-1
 135     latin-iso8859-1
 136     latin-iso8859-2
 137     latin-iso8859-3
 138     latin-iso8859-4
 139     latin-iso8859-9
 140     latin-jisx0201
 141     cyrillic-iso8859-5
 142     greek-iso8859-7
 143     thai-tis620
 144     ;; =mj
 145     ;; =adobe-japan1-0
 146     ;; =adobe-japan1-1
 147     ;; =adobe-japan1-2
 148     ;; =adobe-japan1-3
 149     ;; =adobe-japan1-4
 150     ;; =adobe-japan1-5
 151     ;; =adobe-japan1-6
 152     =jis-x0208
 153     =jis-x0208@1978
 154     =jis-x0208@1983
 155     japanese-jisx0212
 156     chinese-gb2312
 157     =jis-x0208@1990
 158     chinese-cns11643-1
 159     chinese-cns11643-2
 160     chinese-cns11643-3
 161     chinese-cns11643-4
 162     chinese-cns11643-5
 163     chinese-cns11643-6
 164     chinese-cns11643-7
 165     =jis-x0213-1
 166     =jis-x0213-1@2000
 167     =jis-x0213-1@2004
 168     =jis-x0213-2
 169     korean-ksc5601
 170     chinese-isoir165
 171     katakana-jisx0201
 172     hebrew-iso8859-8
 173     chinese-gb12345
 174     latin-viscii
 175     ethiopic-ucs
 176     =big5-cdp
 177     =hanyo-denshi/ja
 178     =hanyo-denshi/jb
 179     =hanyo-denshi/jc
 180     =hanyo-denshi/jd
 181     =hanyo-denshi/ft
 182     =hanyo-denshi/ia
 183     =hanyo-denshi/ib
 184     =hanyo-denshi/hg
 185     =hanyo-denshi/jt
 186     =hanyo-denshi/ks
 187     =hanyo-denshi/tk
 188     ;; ==mj
 189     ;; ==adobe-japan1-0
 190     ;; ==adobe-japan1-1
 191     ;; ==adobe-japan1-2
 192     ;; ==adobe-japan1-3
 193     ;; ==adobe-japan1-4
 194     ;; ==adobe-japan1-5
 195     ;; ==adobe-japan1-6
 196     ==jis-x0208
 197     ==jis-x0213-1
 198     ==jis-x0213-2
 199     ==jis-x0212
 200     ==hanyo-denshi/ja
 201     ==hanyo-denshi/jb
 202     ==hanyo-denshi/jc
 203     ==hanyo-denshi/ft
 204     ==hanyo-denshi/ia
 205     ==hanyo-denshi/ib
 206     ==hanyo-denshi/hg
 207     ==hanyo-denshi/jt
 208     ==hanyo-denshi/ks
 209     ==hanyo-denshi/tk
 210     =gt
 211     =gt-k
 212     =daikanwa
 213     =daijiten
 214     =daikanwa@rev2
 215     =daikanwa@rev1
 216     =daikanwa/+p
 217     =daikanwa/ho
 218     ==daikanwa
 219     ==daijiten
 220     =cbeta
 221     ideograph-hanziku-1
 222     ideograph-hanziku-2
 223     ideograph-hanziku-3
 224     ideograph-hanziku-4
 225     ideograph-hanziku-5
 226     ideograph-hanziku-6
 227     ideograph-hanziku-7
 228     ideograph-hanziku-8
 229     ideograph-hanziku-9
 230     ideograph-hanziku-10
 231     ideograph-hanziku-11
 232     ideograph-hanziku-12
 233     ;; =>>>adobe-japan1-0
 234     ;; =>>>adobe-japan1-1
 235     ;; =>>>adobe-japan1-2
 236     ;; =>>>adobe-japan1-3
 237     ;; =>>>adobe-japan1-4
 238     ;; =>>>adobe-japan1-5
 239     ;; =>>>adobe-japan1-6
 240     ;; =>>>jis-x0208
 241     ;; =>>>jis-x0213-1
 242     ;; =>>>jis-x0213-2
 243     ;; =>>>hanyo-denshi/ja
 244     ;; =>>>hanyo-denshi/jb
 245     ;; =>>>hanyo-denshi/jc
 246     ;; =>>>hanyo-denshi/ft
 247     ;; =>>>hanyo-denshi/ib
 248     ;; =>>>hanyo-denshi/hg
 249     ;; =>>>hanyo-denshi/jt
 250     ;; =>>>hanyo-denshi/ks
 251     ;; =>>>gt
 252     =>>adobe-japan1-0
 253     =>>adobe-japan1-1
 254     =>>adobe-japan1-2
 255     =>>adobe-japan1-3
 256     =>>adobe-japan1-4
 257     =>>adobe-japan1-5
 258     =>>adobe-japan1-6
 259     =>>jis-x0208
 260     =>>jis-x0213-1
 261     =>>jis-x0213-1@2000
 262     =>>jis-x0213-1@2004
 263     =>>jis-x0213-2
 264     =>>jis-x0208@1978
 265     =>>hanyo-denshi/ft
 266     =>>hanyo-denshi/jt
 267     =>>hanyo-denshi/ks
 268     =>>gt
 269     =>>daikanwa
 270     =>>cbeta
 271     =+>jis-x0208
 272     =+>jis-x0213-1
 273     =+>jis-x0213-2
 274     =+>adobe-japan1-0
 275     =+>adobe-japan1-1
 276     =+>adobe-japan1-2
 277     =+>adobe-japan1-3
 278     =+>adobe-japan1-4
 279     =+>adobe-japan1-5
 280     =+>adobe-japan1-6
 281     =+>jis-x0208@1978
 282     =>jis-x0208
 283     =>jis-x0208@1997
 284     =>jis-x0213-1
 285     =>jis-x0213-1@2000
 286     =>jis-x0213-1@2004
 287     =>jis-x0213-2
 288     ==>ucs@bucs
 289     ==>daijiten
 290     ;; =>ucs@hanyo-denshi
 291     =>ucs@iso
 292     =>ucs@unicode
 293     =>ucs@jis
 294     =>ucs@cns
 295     =>ucs@ks
 296     =+>ucs@iso
 297     =+>ucs@unicode
 298     =+>ucs@jis
 299     =+>ucs@jis/1990
 300     =+>ucs@cns
 301     =+>ucs@ks
 302     =>>ucs@iso
 303     =>>ucs@unicode
 304     =>>ucs@jis
 305     =>>ucs@cns
 306     =>>>ucs@iso
 307     =>>>ucs@unicode
 308     ==ucs@iso
 309     ==ucs@unicode
 310     ;; ==ucs@cns
 311     ==gb2312
 312     ==ks-x1001
 313     ==cns11643-1
 314     ==cns11643-2
 315     ==cns11643-3
 316     ==cns11643-4
 317     ==cns11643-5
 318     ==cns11643-6
 319     ==cns11643-7
 320     ==gt
 321     ==jis-x0208@1990
 322     ;; ==jis-x0208@1983
 323     ==jis-x0208@1978
 324     ==gt-k
 325     =ucs@iso
 326     =ucs@unicode
 327     =ucs@cns
 328     ==big5-cdp
 329     ==cbeta
 330     =>>big5-cdp
 331     =>>gt-k
 332     =+>gt
 333     =+>big5-cdp
 334     =>big5-cdp
 335     =>big5-cdp@iwds-1
 336     =>mj
 337     =>hanyo-denshi/tk
 338     =>iwds-1
 339     =big5
 340     =big5-eten
 341     =zinbun-oracle
 342     =>zinbun-oracle
 343     =ruimoku-v6
 344     =>>ruimoku-v6
 345     ==ruimoku-v6
 346     =>ruimoku-v6
 347     =jef-china3
 348     =shinjigen
 349     =>cns11643-5
 350     =>cns11643-7
 351     =>gt
 352     =>gt-k
 353     =>daikanwa
 354     =>daikanwa/ho
 355     =>cbeta
 356     =ucs-var-001
 357     =ucs-var-002
 358     =ucs-var-003
 359     =ucs-var-004
 360     =ucs-var-005
 361     =ucs-var-006
 362     =ucs-var-007
 363     =ucs-var-008
 364     =ucs-var-010
 365     =ucs-itaiji-001
 366     =ucs-itaiji-002
 367     =ucs-itaiji-003
 368     =ucs-itaiji-004
 369     =ucs-itaiji-005
 370     =ucs-itaiji-006
 371     =ucs-itaiji-007
 372     =ucs-itaiji-008
 373     =ucs-itaiji-009
 374     =ucs-itaiji-010
 375     =ucs-itaiji-011
 376     =ucs-itaiji-012
 377     =ucs-itaiji-014
 378     =ucs-itaiji-016
 379     =+>ucs-var-003
 380     =>ucs-itaiji-001
 381     =>ucs-itaiji-002
 382     =>ucs-itaiji-003
 383     =>ucs-itaiji-004
 384     =>ucs-itaiji-005
 385     =>ucs-itaiji-006
 386     =>ucs-itaiji-007
 387     =>ucs-itaiji-008
 388     =>ucs-itaiji-009
 389     =>ucs-itaiji-010
 390     =>ucs-itaiji-011
 391     =>ucs-itaiji-012
 392     =>ucs-itaiji-014
 393     =>ucs-itaiji-016
 394     =big5-cdp-var-001
 395     =big5-cdp-var-002
 396     =big5-cdp-var-003
 397     =big5-cdp-var-004
 398     =big5-cdp-var-005
 399     =big5-cdp-var-007
 400     =big5-cdp-var-008
 401     =big5-cdp-var-010
 402     =big5-cdp-itaiji-001
 403     =big5-cdp-itaiji-002
 404     =big5-cdp-itaiji-003
 405     =big5-cdp-itaiji-004
 406     =big5-cdp-itaiji-005
 407     =>big5-cdp-itaiji-001
 408     =>big5-cdp-itaiji-002
 409     =>big5-cdp-itaiji-003
 410     =>big5-cdp-itaiji-004
 411     =>big5-cdp-itaiji-005
 412     =>big5-cdp-itaiji-006
 413     =>big5-cdp-var-001
 414     =>ucs@iwds-1
 415     =>ucs@cognate
 416     =>ucs@component
 417     =>ucs@iwds-1/normalized
 418     =>ucs-itaiji-001@iwds-1
 419     =>ucs-itaiji-002@iwds-1
 420     =>ucs-itaiji-003@iwds-1
 421     =>ucs-itaiji-006@iwds-1
 422     =>big5-cdp@component
 423     =>big5-cdp@cognate
 424     ==ucs@gb
 425     =ucs@gb
 426     ==ucs-var-002
 427     ==ucs-itaiji-002
 428     =ucs@JP/hanazono
 429     =koseki
 430     =shuowen-jiguge
 431     =>shuowen-jiguge
 432     ===shuowen-jiguge5
 433     ===shuowen-jiguge4
 434     ===chise-hdic-ktb-seal))
 435
 436
 437 ;;; @ char-db formatters
 438 ;;;
 439
 440 (defun char-db-make-char-spec (char)
 441   (let (ret char-spec)
 442     (cond ((characterp char)
 443            (cond ((and (setq ret (encode-char char '=ucs 'defined-only))
 444                        (not (and (<= #xE000 ret)(<= ret #xF8FF))))
 445                   (setq char-spec (list (cons '=ucs ret)))
 446                   (cond ((setq ret (get-char-attribute char 'name))
 447                          (setq char-spec (cons (cons 'name ret) char-spec))
 448                          )
 449                         ((setq ret (get-char-attribute char 'name*))
 450                          (setq char-spec (cons (cons 'name* ret) char-spec))
 451                          ))
 452                   )
 453                  ((encode-char char '=mj 'defined-only)
 454                   (setq char-spec nil)
 455                   (dolist (ccs (charset-list))
 456                     (if (and (or (eq ccs '=mj)
 457                                  ;; (eq (charset-property ccs 'iso-ir) 177)
 458                                  (string-match "=ucs@" (symbol-name ccs))
 459                                  )
 460                              (setq ccs (charset-name ccs))
 461                              (null (assq ccs char-spec))
 462                              (setq ret (encode-char char ccs 'defined-only)))
 463                         (setq char-spec (cons (cons ccs ret) char-spec))))
 464                   )
 465                  ((encode-char char '==mj 'defined-only)
 466                   (setq char-spec nil)
 467                   (dolist (ccs (charset-list))
 468                     (if (and (or (eq ccs '==mj)
 469                                  ;; (eq (charset-property ccs 'iso-ir) 177)
 470                                  (string-match "=ucs@" (symbol-name ccs))
 471                                  )
 472                              (setq ccs (charset-name ccs))
 473                              (null (assq ccs char-spec))
 474                              (setq ret (encode-char char ccs 'defined-only)))
 475                         (setq char-spec (cons (cons ccs ret) char-spec))))
 476                   )
 477                  ((encode-char char '=adobe-japan1 'defined-only)
 478                   (setq char-spec nil)
 479                   (dolist (ccs (charset-list))
 480                     (if (and (or (memq ccs
 481                                        '(=adobe-japan1-0
 482                                          =adobe-japan1-1
 483                                          =adobe-japan1-2
 484                                          =adobe-japan1-3
 485                                          =adobe-japan1-4
 486                                          =adobe-japan1-5
 487                                          =adobe-japan1-6
 488                                          ))
 489                                  ;; (eq (charset-property ccs 'iso-ir) 177)
 490                                  (string-match "=ucs@" (symbol-name ccs))
 491                                  )
 492                              (setq ccs (charset-name ccs))
 493                              (null (assq ccs char-spec))
 494                              (setq ret (encode-char char ccs 'defined-only)))
 495                         (setq char-spec (cons (cons ccs ret) char-spec))))
 496                   )
 497                  ((encode-char char '==adobe-japan1 'defined-only)
 498                   (setq char-spec nil)
 499                   (dolist (ccs (charset-list))
 500                     (if (and (or (memq ccs
 501                                        '(==adobe-japan1-0
 502                                          ==adobe-japan1-1
 503                                          ==adobe-japan1-2
 504                                          ==adobe-japan1-3
 505                                          ==adobe-japan1-4
 506                                          ==adobe-japan1-5
 507                                          ==adobe-japan1-6
 508                                          ))
 509                                  ;; (eq (charset-property ccs 'iso-ir) 177)
 510                                  (string-match "=ucs@" (symbol-name ccs))
 511                                  )
 512                              (setq ccs (charset-name ccs))
 513                              (null (assq ccs char-spec))
 514                              (setq ret (encode-char char ccs 'defined-only)))
 515                         (setq char-spec (cons (cons ccs ret) char-spec))))
 516                   )
 517                  ((setq ret
 518                         (catch 'tag
 519                           (let ((rest char-db-coded-charset-priority-list)
 520                                 ccs)
 521                             (while rest
 522                               (setq ccs (charset-name
 523                                          (find-charset (car rest))))
 524                               (if (setq ret
 525                                         (encode-char char ccs
 526                                                      'defined-only))
 527                                   (throw 'tag (cons ccs ret)))
 528                               (setq rest (cdr rest))))))
 529                   (setq char-spec (list ret))
 530                   (dolist (ccs (delq (car ret) (charset-list)))
 531                     (if (and (or (charset-iso-final-char ccs)
 532                                  (memq ccs
 533                                        '(=daikanwa
 534                                          =daikanwa@rev2
 535                                          ;; =gt-k
 536                                          =jis-x0208@1997
 537                                          ))
 538                                  (and (eq (charset-property ccs 'iso-ir) 177)
 539                                       (not (eq (charset-name ccs) '=ucs@big5)))
 540                                  ;; (string-match "=ucs@" (symbol-name ccs))
 541                                  )
 542                              (setq ccs (charset-name ccs))
 543                              (null (assq ccs char-spec))
 544                              (setq ret (encode-char char ccs 'defined-only)))
 545                         (setq char-spec (cons (cons ccs ret) char-spec))))
 546                   (if (null char-spec)
 547                       (setq char-spec (split-char char)))
 548                   (cond ((setq ret (get-char-attribute char 'name))
 549                          (setq char-spec (cons (cons 'name ret) char-spec))
 550                          )
 551                         ((setq ret (get-char-attribute char 'name*))
 552                          (setq char-spec (cons (cons 'name* ret) char-spec))
 553                          ))
 554                   )
 555                  ((setq ret (get-char-attribute
 556                              char 'ideographic-combination))
 557                   (setq char-spec
 558                         (cons (cons 'ideographic-combination ret)
 559                               char-spec))
 560                   ))
 561            char-spec)
 562           ((consp char)
 563            char))))
 564
 565 (defun char-db-insert-char-spec (char &optional readable column
 566                                       required-features)
 567   (unless column
 568     (setq column (current-column)))
 569   (let (char-spec temp-char)
 570     (setq char-spec (char-db-make-char-spec char))
 571     (unless (or (characterp char) ; char
 572                 (condition-case nil
 573                     (setq char (find-char char-spec))
 574                   (error nil)))
 575       ;; define temporary character
 576       ;;   Current implementation is dirty.
 577       (setq temp-char (define-char (cons '(ideograph-daikanwa . 0)
 578                                          char-spec)))
 579       (remove-char-attribute temp-char 'ideograph-daikanwa)
 580       (setq char temp-char))
 581     (insert-char-attributes char
 582                             readable
 583                             (union (mapcar #'car char-spec)
 584                                    required-features)
 585                             nil 'for-sub-node)
 586     (when temp-char
 587       ;; undefine temporary character
 588       ;;   Current implementation is dirty.
 589       (setq char-spec (char-attribute-alist temp-char))
 590       (while char-spec
 591         (remove-char-attribute temp-char (car (car char-spec)))
 592         (setq char-spec (cdr char-spec))))))
 593
 594 (defun char-db-insert-alist (alist &optional readable column)
 595   (unless column
 596     (setq column (current-column)))
 597   (let ((line-breaking
 598          (concat "\n" (make-string (1+ column) ?\ )))
 599         name value
 600         ret al ; cal
 601         key
 602         lbs cell rest separator)
 603     (insert "(")
 604     (while alist
 605       (setq name (car (car alist))
 606             value (cdr (car alist)))
 607       (cond ((eq name 'char)
 608              (insert "(char . ")
 609              (if (setq ret (condition-case nil
 610                                (find-char value)
 611                              (error nil)))
 612                  (progn
 613                    (setq al nil
 614                          ;; cal nil
 615                          )
 616                    (while value
 617                      (setq key (car (car value)))
 618                      ;; (if (find-charset key)
 619                      ;;     (setq cal (cons key cal))
 620                      (setq al (cons key al))
 621                      ;; )
 622                      (setq value (cdr value)))
 623                    (insert-char-attributes ret
 624                                            readable
 625                                            (or al 'none) ; cal
 626                                            nil 'for-sub-node))
 627                (insert (prin1-to-string value)))
 628              (insert ")")
 629              (insert line-breaking))
 630             ((consp value)
 631              (insert (format "(%-18s " name))
 632              (setq lbs (concat "\n" (make-string (current-column) ?\ )))
 633              (while (consp value)
 634                (setq cell (car value))
 635                (if (and (consp cell)
 636                         (consp (car cell))
 637                         (setq ret (condition-case nil
 638                                       (find-char cell)
 639                                     (error nil)))
 640                         )
 641                    (progn
 642                      (setq rest cell
 643                            al nil
 644                            ;; cal nil
 645                            )
 646                      (while rest
 647                        (setq key (car (car rest)))
 648                        ;; (if (find-charset key)
 649                        ;;     (setq cal (cons key cal))
 650                        (setq al (cons key al))
 651                        ;; )
 652                        (setq rest (cdr rest)))
 653                      (if separator
 654                          (insert lbs))
 655                      (insert-char-attributes ret
 656                                              readable
 657                                              al ; cal
 658                                              nil 'for-sub-node)
 659                      (setq separator lbs))
 660                  (if separator
 661                      (insert separator))
 662                  (insert (prin1-to-string cell))
 663                  (setq separator " "))
 664                (setq value (cdr value)))
 665              (insert ")")
 666              (insert line-breaking))
 667             (t
 668              (insert (format "(%-18s . %S)%s"
 669                              name value
 670                              line-breaking))))
 671       (setq alist (cdr alist))))
 672   (insert ")"))
 673
 674 (defun char-db-insert-char-reference (plist &optional readable column)
 675   (unless column
 676     (setq column (current-column)))
 677   (let ((line-breaking
 678          (concat "\n" (make-string (1+ column) ?\ )))
 679         (separator "")
 680         name value)
 681     (insert "(")
 682     (while plist
 683       (setq name (pop plist))
 684       (setq value (pop plist))
 685       (cond ((eq name :char)
 686              (insert separator)
 687              (insert ":char\t")
 688              (cond ((numberp value)
 689                     (setq value (decode-char '=ucs value)))
 690                    ;; ((consp value)
 691                    ;;  (setq value (or (find-char value)
 692                    ;;                  value)))
 693                    )
 694              (char-db-insert-char-spec value readable)
 695              (insert line-breaking)
 696              (setq separator ""))
 697             ((eq name :radical)
 698              (insert (format "%s%s\t%d ; %c%s"
 699                              separator
 700                              name value
 701                              (ideographic-radical value)
 702                              line-breaking))
 703              (setq separator ""))
 704             (t
 705              (insert (format "%s%s\t%S" separator name value))
 706              (setq separator line-breaking)))
 707       ))
 708   (insert ")"))
 709
 710 (defun char-db-decode-isolated-char (ccs code-point)
 711   (let (ret)
 712     (setq ret
 713           (cond ((eq ccs 'arabic-iso8859-6)
 714                  (decode-char ccs code-point))
 715                 ;; ((eq ccs '=gt)
 716                 ;;  (decode-builtin-char '==gt code-point))
 717                 ((and (memq ccs '(=gt-pj-1
 718                                   =gt-pj-2
 719                                   =gt-pj-3
 720                                   =gt-pj-4
 721                                   =gt-pj-5
 722                                   =gt-pj-6
 723                                   =gt-pj-7
 724                                   =gt-pj-8
 725                                   =gt-pj-9
 726                                   =gt-pj-10
 727                                   =gt-pj-11))
 728                       (setq ret (decode-char ccs code-point))
 729                       (setq ret (encode-char ret '=gt 'defined-only)))
 730                  (decode-builtin-char '=gt ret))
 731                 (t
 732                  (decode-builtin-char ccs code-point))))
 733     (cond ((null ret)
 734            (or (decode-char ccs code-point)
 735                (define-char (list (cons ccs code-point)))))
 736           ((and (<= 0 (char-int ret))
 737                 (<= (char-int ret) #x1F))
 738            (decode-char '=ucs (+ #x2400 (char-int ret))))
 739           ((= (char-int ret) #x7F)
 740            ?\u2421)
 741           (t ret))))
 742
 743 (defvar char-db-convert-obsolete-format t)
 744
 745 (defun char-db-insert-ccs-feature (name value line-breaking)
 746   (cond
 747    ((integerp value)
 748     (insert
 749      (format
 750       (cond
 751        ((memq name '(=>iwds-1
 752                      ==shinjigen
 753                      =shinjigen
 754                      =shinjigen@1ed ==shinjigen@1ed
 755                      =shinjigen@rev ==shinjigen@rev
 756                      =shinjigen/+p@rev ==shinjigen/+p@rev
 757                      ===daikanwa/ho ==daikanwa/ho
 758                      =daikanwa/ho =>>daikanwa/ho =>daikanwa/ho))
 759         "(%-18s .  %04d)\t; %c")
 760        ((eq name '=shinjigen@1ed/24pr)
 761         "(%-18s . %04d)\t; %c")
 762        ((or
 763          (memq name
 764                '(===daikanwa
 765                  ==daikanwa =daikanwa =>>daikanwa =>daikanwa
 766                  =daikanwa@rev1 =daikanwa@rev2
 767                  =daikanwa/+p ==daikanwa/+p ===daikanwa/+p
 768                  =>>daikanwa/+p
 769                  =daikanwa/+2p =>>daikanwa/+2p
 770                  =gt ==gt ===gt
 771                  =>>gt =+>gt =>gt
 772                  =gt-k ==gt-k ===gt-k
 773                  =>>gt-k =>gt-k
 774                  =adobe-japan1-0 ==adobe-japan1-0 ===adobe-japan1-0
 775                  =adobe-japan1-1 ==adobe-japan1-1 ===adobe-japan1-1
 776                  =adobe-japan1-2 ==adobe-japan1-2 ===adobe-japan1-2
 777                  =adobe-japan1-3 ==adobe-japan1-3 ===adobe-japan1-3
 778                  =adobe-japan1-4 ==adobe-japan1-4 ===adobe-japan1-4
 779                  =adobe-japan1-5 ==adobe-japan1-5 ===adobe-japan1-5
 780                  =adobe-japan1-6 ==adobe-japan1-6 ===adobe-japan1-6
 781                  =>>adobe-japan1-0 =+>adobe-japan1-0
 782                  =>>adobe-japan1-1 =+>adobe-japan1-1
 783                  =>>adobe-japan1-2 =+>adobe-japan1-2
 784                  =>>adobe-japan1-3 =+>adobe-japan1-3
 785                  =>>adobe-japan1-4 =+>adobe-japan1-4
 786                  =>>adobe-japan1-5 =+>adobe-japan1-5
 787                  =>>adobe-japan1-6 =+>adobe-japan1-6
 788                  =>cbeta =cbeta =>>cbeta ==cbeta ===cbeta
 789                  =zinbun-oracle =>zinbun-oracle
 790                  =daijiten ==daijiten ===daijiten ==>daijiten
 791                  ===hng-jou ===hng-keg ===hng-dng ===hng-mam
 792                  ===hng-drt ===hng-kgk ===hng-myz ===hng-kda
 793                  ===hng-khi ===hng-khm ===hng-fhs ===hng-hok
 794                  ===hng-kyd ===hng-sok
 795                  ===hng-yhk ===hng-kak ===hng-kar ===hng-kae
 796                  ===hng-sys ===hng-tsu ===hng-tzj
 797                  ===hng-hos ===hng-kkh ===hng-nak ===hng-jhk
 798                  ===hng-hod ===hng-gok ===hng-ink ===hng-nto
 799                  ===hng-nkm ===hng-k24 ===hng-ini ===hng-nkk
 800                  ===hng-kcc ===hng-kcj ===hng-kbk ===hng-sik
 801                  ===hng-skk ===hng-kyu ===hng-ksk ===hng-wan
 802                  ===hng-okd ===hng-wad ===hng-kmi
 803                  ===hng-zkd ===hng-doh ===hng-jyu ===hng-tzs
 804                  ===hng-sai ===hng-kad ===hng-kss
 805                  ===hng-kyo ===hng-ykk ===hng-saa
 806                  ===hng-sab ===hng-wks ===hng-wke ===hng-smk
 807                  ===hng-sgs ===hng-sts ===hng-kkd
 808                  =>shuowen-jiguge =shuowen-jiguge
 809                  ===shuowen-jiguge4 ===shuowen-jiguge5 ===shuowen-jiguge-A30))
 810          ;; (string-match "^=adobe-" (symbol-name name))
 811          )
 812         "(%-18s . %05d)\t; %c")
 813        ((memq name '(=hanyo-denshi/ks
 814                      ==hanyo-denshi/ks ===hanyo-denshi/ks
 815                      =>>hanyo-denshi/ks
 816                      =koseki ==koseki
 817                      =mj ==mj ===mj =>>mj =>mj
 818                      =zihai mojikyo))
 819         "(%-18s . %06d)\t; %c")
 820        ((memq name '(=hanyo-denshi/tk =>hanyo-denshi/tk ==hanyo-denshi/tk))
 821         "(%-18s . %08d)\t; %c")
 822        ((>= (charset-dimension name) 2)
 823         "(%-18s . #x%04X)\t; %c")
 824        (t
 825         "(%-18s . #x%02X)\t; %c"))
 826       name
 827       (if (= (charset-iso-graphic-plane name) 1)
 828           (logior value
 829                   (cond ((= (charset-dimension name) 1)
 830                          #x80)
 831                         ((= (charset-dimension name) 2)
 832                          #x8080)
 833                         ((= (charset-dimension name) 3)
 834                          #x808080)
 835                         (t 0)))
 836         value)
 837       (char-db-decode-isolated-char name value)))
 838     (if (and (= (charset-chars name) 94)
 839              (= (charset-dimension name) 2))
 840         (insert (format " [%02d-%02d]"
 841                         (- (lsh value -8) 32)
 842                         (- (logand value 255) 32))))
 843     )
 844    (t
 845     (insert (format "(%-18s . %s)" name value))
 846     ))
 847   (insert line-breaking))
 848
 849 (defun char-db-insert-relation-feature (char name value line-breaking
 850                                              ccss readable)
 851   (insert (format "(%-18s%s " name line-breaking))
 852   (let ((lbs (concat "\n" (make-string (current-column) ?\ )))
 853         separator cell sources required-features
 854         ret)
 855     (while (consp value)
 856       (setq cell (car value))
 857       (if (integerp cell)
 858           (setq cell (decode-char '=ucs cell)))
 859       (cond
 860        ((eq name '->subsumptive)
 861         (when (or (not (some (lambda (atr)
 862                                (get-char-attribute cell atr))
 863                              char-db-ignored-attributes))
 864                   (some (lambda (ccs)
 865                           (encode-char cell ccs 'defined-only))
 866                         ccss))
 867           (if separator
 868               (insert lbs))
 869           (let ((char-db-ignored-attributes
 870                  (cons '<-subsumptive
 871                        char-db-ignored-attributes)))
 872             (insert-char-attributes cell readable nil nil 'for-sub-node))
 873           (setq separator lbs))
 874         )
 875        ((characterp cell)
 876         (setq sources
 877               (get-char-attribute
 878                char (intern (format "%s*sources" name))))
 879         (setq required-features nil)
 880         (dolist (source sources)
 881           (cond
 882            ((memq source '(JP
 883                            JP/Jouyou
 884                            shinjigen shinjigen@1ed shinjigen@rev))
 885             (setq required-features
 886                   (union required-features
 887                          '(=jis-x0208
 888                            =jis-x0208@1990
 889                            =jis-x0213-1@2000
 890                            =jis-x0213-1@2004
 891                            =jis-x0213-2
 892                            =jis-x0212
 893                            =jis-x0208@1983
 894                            =jis-x0208@1978
 895                            =shinjigen
 896                            =shinjigen@1ed
 897                            =shinjigen@rev
 898                            =shinjigen/+p@rev))))
 899            ((eq source 'CN)
 900             (setq required-features
 901                   (union required-features
 902                          '(=gb2312
 903                            =gb12345
 904                            =iso-ir165)))))
 905           (cond
 906            ((find-charset (setq ret (intern (format "=%s" source))))
 907             (setq required-features
 908                   (cons ret required-features)))
 909            (t (setq required-features
 910                     (cons source required-features)))))
 911         (cond ((string-match "@JP" (symbol-name name))
 912                (setq required-features
 913                      (union required-features
 914                             '(=jis-x0208
 915                               =jis-x0208@1990
 916                               =jis-x0213-1-2000
 917                               =jis-x0213-2-2000
 918                               =jis-x0212
 919                               =jis-x0208@1983
 920                               =jis-x0208@1978))))
 921               ((string-match "@CN" (symbol-name name))
 922                (setq required-features
 923                      (union required-features
 924                             '(=gb2312
 925                               =gb12345
 926                               =iso-ir165)))))
 927         (if separator
 928             (insert lbs))
 929         (if readable
 930             (insert (format "%S" cell))
 931           (char-db-insert-char-spec cell readable
 932                                     nil
 933                                     required-features))
 934         (setq separator lbs))
 935        ((consp cell)
 936         (if separator
 937             (insert lbs))
 938         (if (consp (car cell))
 939             (char-db-insert-char-spec cell readable)
 940           (char-db-insert-char-reference cell readable))
 941         (setq separator lbs))
 942        (t
 943         (if separator
 944             (insert separator))
 945         (insert (prin1-to-string cell))
 946         (setq separator " ")))
 947       (setq value (cdr value)))
 948     (insert ")")
 949     (insert line-breaking)))
 950
 951 (defun insert-char-attributes (char &optional readable attributes column
 952                                     for-sub-node)
 953   (unless column
 954     (setq column (current-column)))
 955   (let (name value ; has-long-ccs-name
 956         rest
 957         radical strokes
 958         (line-breaking
 959          (concat "\n" (make-string (1+ column) ?\ )))
 960         lbs cell separator ret
 961         key al cal
 962         dest-ccss ; sources required-features
 963         ccss)
 964     (let (atr-d)
 965       (setq attributes
 966             (sort (if attributes
 967                       (if (consp attributes)
 968                           (progn
 969                             (dolist (name attributes)
 970                               (unless (memq name char-db-ignored-attributes)
 971                                 (if (find-charset name)
 972                                     (push name ccss))
 973                                 (push name atr-d)))
 974                             atr-d))
 975                     (dolist (name (char-attribute-list))
 976                       (unless (memq name char-db-ignored-attributes)
 977                         (if (find-charset name)
 978                             (push name ccss))
 979                         (push name atr-d)))
 980                     atr-d)
 981                   #'char-attribute-name<)))
 982     (insert "(")
 983     (when (memq '<-subsumptive attributes)
 984       (when (or readable (not for-sub-node))
 985         (when (setq value (get-char-attribute char '<-subsumptive))
 986           (char-db-insert-relation-feature char '<-subsumptive value
 987                                            line-breaking
 988                                            ccss readable)))
 989       (setq attributes (delq '<-subsumptive attributes)))
 990     (when (and (memq '<-denotational attributes)
 991                (setq value (get-char-attribute char '<-denotational)))
 992       (char-db-insert-relation-feature char '<-denotational value
 993                                        line-breaking
 994                                        ccss readable)
 995       (setq attributes (delq '<-denotational attributes)))
 996     (when (and (memq '<-denotational@component attributes)
 997                (setq value (get-char-attribute char '<-denotational@component)))
 998       (char-db-insert-relation-feature char '<-denotational@component value
 999                                        line-breaking
1000                                        ccss readable)
1001       (setq attributes (delq '<-denotational@component attributes)))
1002     (when (and (memq '<-denotational@usage attributes)
1003                (setq value (get-char-attribute char '<-denotational@usage)))
1004       (char-db-insert-relation-feature char '<-denotational@usage value
1005                                        line-breaking
1006                                        ccss readable)
1007       (setq attributes (delq '<-denotational@usage attributes)))
1008     (when (and (memq 'name attributes)
1009                (setq value (get-char-attribute char 'name)))
1010       (insert (format
1011                (if (> (+ (current-column) (length value)) 48)
1012                    "(name . %S)%s"
1013                  "(name               . %S)%s")
1014                value line-breaking))
1015       (setq attributes (delq 'name attributes))
1016       )
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 'script attributes)
1027                (setq value (get-char-attribute char 'script)))
1028       (insert (format "(script\t\t%s)%s"
1029                       (mapconcat (function prin1-to-string)
1030                                  value " ")
1031                       line-breaking))
1032       (setq attributes (delq 'script attributes))
1033       )
1034     (dolist (name '(=>ucs =>ucs*))
1035       (when (and (memq name attributes)
1036                  (setq value (get-char-attribute char name)))
1037         (insert (format "(%-18s . #x%04X)\t; %c%s"
1038                         name value (decode-char '=ucs value)
1039                         line-breaking))
1040         (setq attributes (delq name attributes))))
1041     (dolist (name '(=>ucs@gb =>ucs@big5))
1042       (when (and (memq name attributes)
1043                  (setq value (get-char-attribute char name)))
1044         (insert (format "(%-18s . #x%04X)\t; %c%s"
1045                         name value
1046                         (decode-char (intern
1047                                       (concat "="
1048                                               (substring
1049                                                (symbol-name name) 2)))
1050                                      value)
1051                         line-breaking))
1052         (setq attributes (delq name attributes))
1053         ))
1054     ;; (dolist (name '(=>daikanwa))
1055     ;;   (when (and (memq name attributes)
1056     ;;              (setq value (get-char-attribute char name)))
1057     ;;     (insert
1058     ;;      (if (integerp value)
1059     ;;          (format "(%-18s . %05d)\t; %c%s"
1060     ;;                  name value (decode-char '=daikanwa value)
1061     ;;                  line-breaking)
1062     ;;        (format "(%-18s %s)\t; %c%s"
1063     ;;                name
1064     ;;                (mapconcat (function prin1-to-string)
1065     ;;                           value " ")
1066     ;;                (char-representative-of-daikanwa char)
1067     ;;                line-breaking)))
1068     ;;     (setq attributes (delq name attributes))))
1069     (when (and (memq 'general-category attributes)
1070                (setq value (get-char-attribute char 'general-category)))
1071       (insert (format
1072                "(general-category\t%s) ; %s%s"
1073                (mapconcat (lambda (cell)
1074                             (format "%S" cell))
1075                           value " ")
1076                (cond ((rassoc value unidata-normative-category-alist)
1077                       "Normative Category")
1078                      ((rassoc value unidata-informative-category-alist)
1079                       "Informative Category")
1080                      (t
1081                       "Unknown Category"))
1082                line-breaking))
1083       (setq attributes (delq 'general-category attributes))
1084       )
1085     (when (and (memq 'bidi-category attributes)
1086                (setq value (get-char-attribute char 'bidi-category)))
1087       (insert (format "(bidi-category\t. %S)%s"
1088                       value
1089                       line-breaking))
1090       (setq attributes (delq 'bidi-category attributes))
1091       )
1092     (unless (or (not (memq 'mirrored attributes))
1093                 (eq (setq value (get-char-attribute char 'mirrored 'empty))
1094                     'empty))
1095       (insert (format "(mirrored\t\t. %S)%s"
1096                       value
1097                       line-breaking))
1098       (setq attributes (delq 'mirrored attributes))
1099       )
1100     (cond
1101      ((and (memq 'decimal-digit-value attributes)
1102            (setq value (get-char-attribute char 'decimal-digit-value)))
1103       (insert (format "(decimal-digit-value . %S)%s"
1104                       value
1105                       line-breaking))
1106       (setq attributes (delq 'decimal-digit-value attributes))
1107       (when (and (memq 'digit-value attributes)
1108                  (setq value (get-char-attribute char 'digit-value)))
1109         (insert (format "(digit-value\t . %S)%s"
1110                         value
1111                         line-breaking))
1112         (setq attributes (delq 'digit-value attributes))
1113         )
1114       (when (and (memq 'numeric-value attributes)
1115                  (setq value (get-char-attribute char 'numeric-value)))
1116         (insert (format "(numeric-value\t . %S)%s"
1117                         value
1118                         line-breaking))
1119         (setq attributes (delq 'numeric-value attributes))
1120         )
1121       )
1122      (t
1123       (when (and (memq 'digit-value attributes)
1124                  (setq value (get-char-attribute char 'digit-value)))
1125         (insert (format "(digit-value\t. %S)%s"
1126                         value
1127                         line-breaking))
1128         (setq attributes (delq 'digit-value attributes))
1129         )
1130       (when (and (memq 'numeric-value attributes)
1131                  (setq value (get-char-attribute char 'numeric-value)))
1132         (insert (format "(numeric-value\t. %S)%s"
1133                         value
1134                         line-breaking))
1135         (setq attributes (delq 'numeric-value attributes))
1136         )))
1137     (when (and (memq 'iso-10646-comment attributes)
1138                (setq value (get-char-attribute char 'iso-10646-comment)))
1139       (insert (format "(iso-10646-comment\t. %S)%s"
1140                       value
1141                       line-breaking))
1142       (setq attributes (delq 'iso-10646-comment attributes))
1143       )
1144     (when (and (memq 'morohashi-daikanwa attributes)
1145                (setq value (get-char-attribute char 'morohashi-daikanwa)))
1146       (insert (format "(morohashi-daikanwa\t%s)%s"
1147                       (mapconcat (function prin1-to-string) value " ")
1148                       line-breaking))
1149       (setq attributes (delq 'morohashi-daikanwa attributes))
1150       )
1151     (setq radical nil
1152           strokes nil)
1153     (when (and (memq 'ideographic-radical attributes)
1154                (setq value (get-char-attribute char 'ideographic-radical)))
1155       (setq radical value)
1156       (insert (format "(ideographic-radical . %S)\t; %c%s"
1157                       radical
1158                       (ideographic-radical radical)
1159                       line-breaking))
1160       (setq attributes (delq 'ideographic-radical attributes))
1161       )
1162     (when (and (memq 'shuowen-radical attributes)
1163                (setq value (get-char-attribute char 'shuowen-radical)))
1164       (insert (format "(shuowen-radical\t. %S)\t; %c%s"
1165                       value
1166                       (shuowen-radical value)
1167                       line-breaking))
1168       (setq attributes (delq 'shuowen-radical attributes))
1169       )
1170     (let (key)
1171       (dolist (domain
1172                (append
1173                 char-db-feature-domains
1174                 (let (dest domain)
1175                   (dolist (feature (char-attribute-list))
1176                     (setq feature (symbol-name feature))
1177                     (when (string-match
1178                            "\\(radical\\|strokes\\)@\\([^@*]+\\)\\(\\*\\|$\\)"
1179                            feature)
1180                       (setq domain (intern (match-string 2 feature)))
1181                      (unless (memq domain dest)
1182                        (setq dest (cons domain dest)))))
1183                   (sort dest #'string<))))
1184         (setq key (intern (format "%s@%s" 'ideographic-radical domain)))
1185         (when (and (memq key attributes)
1186                    (setq value (get-char-attribute char key)))
1187           (setq radical value)
1188           (insert (format "(%s . %S)\t; %c%s"
1189                           key
1190                           radical
1191                           (ideographic-radical radical)
1192                           line-breaking))
1193           (setq attributes (delq key attributes))
1194           )
1195         (setq key (intern (format "%s@%s" 'ideographic-strokes domain)))
1196         (when (and (memq key attributes)
1197                    (setq value (get-char-attribute char key)))
1198           (setq strokes value)
1199           (insert (format "(%s . %S)%s"
1200                           key
1201                           strokes
1202                           line-breaking))
1203           (setq attributes (delq key attributes))
1204           )
1205         (setq key (intern (format "%s@%s" 'total-strokes domain)))
1206         (when (and (memq key attributes)
1207                    (setq value (get-char-attribute char key)))
1208           (insert (format "(%s       . %S)%s"
1209                           key
1210                           value
1211                           line-breaking))
1212           (setq attributes (delq key attributes))
1213           )
1214         (dolist (feature '(ideographic-radical
1215                            ideographic-strokes
1216                            total-strokes))
1217           (setq key (intern (format "%s@%s*sources" feature domain)))
1218           (when (and (memq key attributes)
1219                      (setq value (get-char-attribute char key)))
1220             (insert (format "(%s%s" key line-breaking))
1221             (dolist (cell value)
1222               (insert (format " %s" cell)))
1223             (insert ")")
1224             (insert line-breaking)
1225             (setq attributes (delq key attributes))
1226             ))
1227         ))
1228     (when (and (memq 'ideographic-strokes attributes)
1229                (setq value (get-char-attribute char 'ideographic-strokes)))
1230       (setq strokes value)
1231       (insert (format "(ideographic-strokes . %S)%s"
1232                       strokes
1233                       line-breaking))
1234       (setq attributes (delq 'ideographic-strokes attributes))
1235       )
1236     (when (and (memq 'kangxi-radical attributes)
1237                (setq value (get-char-attribute char 'kangxi-radical)))
1238       (unless (eq value radical)
1239         (insert (format "(kangxi-radical\t . %S)\t; %c%s"
1240                         value
1241                         (ideographic-radical value)
1242                         line-breaking))
1243         (or radical
1244             (setq radical value)))
1245       (setq attributes (delq 'kangxi-radical attributes))
1246       )
1247     (when (and (memq 'kangxi-strokes attributes)
1248                (setq value (get-char-attribute char 'kangxi-strokes)))
1249       (unless (eq value strokes)
1250         (insert (format "(kangxi-strokes\t . %S)%s"
1251                         value
1252                         line-breaking))
1253         (or strokes
1254             (setq strokes value)))
1255       (setq attributes (delq 'kangxi-strokes attributes))
1256       )
1257     ;; (when (and (memq 'japanese-radical attributes)
1258     ;;            (setq value (get-char-attribute char 'japanese-radical)))
1259     ;;   (unless (eq value radical)
1260     ;;     (insert (format "(japanese-radical\t . %S)\t; %c%s"
1261     ;;                     value
1262     ;;                     (ideographic-radical value)
1263     ;;                     line-breaking))
1264     ;;     (or radical
1265     ;;         (setq radical value)))
1266     ;;   (setq attributes (delq 'japanese-radical attributes))
1267     ;;   )
1268     (when (and (memq 'japanese-strokes attributes)
1269                (setq value (get-char-attribute char 'japanese-strokes)))
1270       (unless (eq value strokes)
1271         (insert (format "(japanese-strokes\t . %S)%s"
1272                         value
1273                         line-breaking))
1274         (or strokes
1275             (setq strokes value)))
1276       (setq attributes (delq 'japanese-strokes attributes))
1277       )
1278     (when (and (memq 'cns-radical attributes)
1279                (setq value (get-char-attribute char 'cns-radical)))
1280       (insert (format "(cns-radical\t . %S)\t; %c%s"
1281                       value
1282                       (ideographic-radical value)
1283                       line-breaking))
1284       (setq attributes (delq 'cns-radical attributes))
1285       )
1286     (when (and (memq 'cns-strokes attributes)
1287                (setq value (get-char-attribute char 'cns-strokes)))
1288       (unless (eq value strokes)
1289         (insert (format "(cns-strokes\t . %S)%s"
1290                         value
1291                         line-breaking))
1292         (or strokes
1293             (setq strokes value)))
1294       (setq attributes (delq 'cns-strokes attributes))
1295       )
1296     ;; (when (and (memq 'ideographic- attributes)
1297     ;;            (setq value (get-char-attribute char 'ideographic-)))
1298     ;;   (insert "(ideographic-       ")
1299     ;;   (setq lbs (concat "\n" (make-string (current-column) ?\ ))
1300     ;;         separator nil)
1301     ;;   (while (consp value)
1302     ;;     (setq cell (car value))
1303     ;;     (if (integerp cell)
1304     ;;         (setq cell (decode-char '=ucs cell)))
1305     ;;     (cond ((characterp cell)
1306     ;;            (if separator
1307     ;;                (insert lbs))
1308     ;;            (if readable
1309     ;;                (insert (format "%S" cell))
1310     ;;              (char-db-insert-char-spec cell readable))
1311     ;;            (setq separator lbs))
1312     ;;           ((consp cell)
1313     ;;            (if separator
1314     ;;                (insert lbs))
1315     ;;            (if (consp (car cell))
1316     ;;                (char-db-insert-char-spec cell readable)
1317     ;;              (char-db-insert-char-reference cell readable))
1318     ;;            (setq separator lbs))
1319     ;;           (t
1320     ;;            (if separator
1321     ;;                (insert separator))
1322     ;;            (insert (prin1-to-string cell))
1323     ;;            (setq separator " ")))
1324     ;;     (setq value (cdr value)))
1325     ;;   (insert ")")
1326     ;;   (insert line-breaking)
1327     ;;   (setq attributes (delq 'ideographic- attributes)))
1328     (when (and (memq 'total-strokes attributes)
1329                (setq value (get-char-attribute char 'total-strokes)))
1330       (insert (format "(total-strokes       . %S)%s"
1331                       value
1332                       line-breaking))
1333       (setq attributes (delq 'total-strokes attributes))
1334       )
1335     ;; (when (and (memq '->ideograph attributes)
1336     ;;            (setq value (get-char-attribute char '->ideograph)))
1337     ;;   (insert (format "(->ideograph\t%s)%s"
1338     ;;                   (mapconcat (lambda (code)
1339     ;;                                (cond ((symbolp code)
1340     ;;                                       (symbol-name code))
1341     ;;                                      ((integerp code)
1342     ;;                                       (format "#x%04X" code))
1343     ;;                                      (t
1344     ;;                                       (format "%s %S"
1345     ;;                                               line-breaking code))))
1346     ;;                              value " ")
1347     ;;                   line-breaking))
1348     ;;   (setq attributes (delq '->ideograph attributes))
1349     ;;   )
1350     ;; (when (and (memq '->decomposition attributes)
1351     ;;            (setq value (get-char-attribute char '->decomposition)))
1352     ;;   (insert (format "(->decomposition\t%s)%s"
1353     ;;                   (mapconcat (lambda (code)
1354     ;;                                (cond ((symbolp code)
1355     ;;                                       (symbol-name code))
1356     ;;                                      ((characterp code)
1357     ;;                                       (if readable
1358     ;;                                           (format "%S" code)
1359     ;;                                         (format "#x%04X"
1360     ;;                                                 (char-int code))
1361     ;;                                         ))
1362     ;;                                      ((integerp code)
1363     ;;                                       (format "#x%04X" code))
1364     ;;                                      (t
1365     ;;                                       (format "%s%S" line-breaking code))))
1366     ;;                              value " ")
1367     ;;                   line-breaking))
1368     ;;   (setq attributes (delq '->decomposition attributes))
1369     ;;   )
1370     (if (equal (get-char-attribute char '->titlecase)
1371                (get-char-attribute char '->uppercase))
1372         (setq attributes (delq '->titlecase attributes)))
1373     (when (and (memq '->mojikyo attributes)
1374                (setq value (get-char-attribute char '->mojikyo)))
1375       (insert (format "(->mojikyo\t\t. %06d)\t; %c%s"
1376                       value (decode-char 'mojikyo value)
1377                       line-breaking))
1378       (setq attributes (delq '->mojikyo attributes))
1379       )
1380     (unless readable
1381       (dolist (ignored '(composition
1382                          ->denotational <-subsumptive ->ucs-unified
1383                          ;; ->ideographic-component-forms
1384                          ))
1385         (setq attributes (delq ignored attributes))))
1386     (while attributes
1387       (setq name (car attributes))
1388       (unless (eq (setq value (get-char-attribute char name 'value-is-empty))
1389                   'value-is-empty)
1390         (cond ((setq ret (find-charset name))
1391                (setq name (charset-name ret))
1392                (when (not (memq name dest-ccss))
1393                  (setq dest-ccss (cons name dest-ccss))
1394                  (char-db-insert-ccs-feature name value line-breaking))
1395                )
1396               ((string-match "^=>ucs@" (symbol-name name))
1397                (insert (format "(%-18s . #x%04X)\t; %c%s"
1398                                name value (decode-char '=ucs value)
1399                                line-breaking))
1400                )
1401               ((eq name 'jisx0208-1978/4X)
1402                (insert (format "(%-18s . #x%04X)%s"
1403                                name value
1404                                line-breaking))
1405                )
1406               ((and
1407                 (not readable)
1408                 (not (eq name '->subsumptive))
1409                 (not (eq name '->uppercase))
1410                 (not (eq name '->lowercase))
1411                 (not (eq name '->titlecase))
1412                 (not (eq name '->canonical))
1413                 (not (eq name '->Bopomofo))
1414                 (not (eq name '->mistakable))
1415                 (not (eq name '->ideographic-variants))
1416                 (null (get-char-attribute
1417                        char (intern (format "%s*sources" name))))
1418                 (not (string-match "\\*sources$" (symbol-name name)))
1419                 (null (get-char-attribute
1420                        char (intern (format "%s*note" name))))
1421                 (not (string-match "\\*note$" (symbol-name name)))
1422                 (or (eq name '<-identical)
1423                     (eq name '<-uppercase)
1424                     (eq name '<-lowercase)
1425                     (eq name '<-titlecase)
1426                     (eq name '<-canonical)
1427                     (eq name '<-ideographic-variants)
1428                     ;; (eq name '<-synonyms)
1429                     (string-match "^<-synonyms" (symbol-name name))
1430                     (eq name '<-mistakable)
1431                     (when (string-match "^->" (symbol-name name))
1432                       (cond
1433                        ((string-match "^->fullwidth" (symbol-name name))
1434                         (not (and (consp value)
1435                                   (characterp (car value))
1436                                   (encode-char
1437                                    (car value) '=ucs 'defined-only)))
1438                         )
1439                        (t)))
1440                     ))
1441                )
1442               ((or (eq name 'ideographic-structure)
1443                    (eq name 'ideographic-structure@apparent)
1444                    (eq name 'ideographic-structure@apparent/leftmost)
1445                    (eq name 'ideographic-structure@apparent/rightmost)
1446                    (eq name 'ideographic-combination)
1447                    ;; (eq name 'ideographic-)
1448                    (eq name '=decomposition)
1449                    (char-feature-base-name= '=decomposition name)
1450                    (char-feature-base-name= '=>decomposition name)
1451                    ;; (string-match "^=>*decomposition\\(@[^*]+\\)?$"
1452                    ;;               (symbol-name name))
1453                    (string-match "^\\(->\\|<-\\)[^*]*$" (symbol-name name))
1454                    (string-match "^\\(->\\|<-\\)[^*]*\\*sources$"
1455                                  (symbol-name name))
1456                    )
1457                (char-db-insert-relation-feature char name value
1458                                                 line-breaking
1459                                                 ccss readable))
1460               ;; ((memq name '(ideograph=
1461               ;;               original-ideograph-of
1462               ;;               ancient-ideograph-of
1463               ;;               vulgar-ideograph-of
1464               ;;               wrong-ideograph-of
1465               ;;               ;; simplified-ideograph-of
1466               ;;               ideographic-variants
1467               ;;               ;; ideographic-different-form-of
1468               ;;               ))
1469               ;;  (insert (format "(%-18s%s " name line-breaking))
1470               ;;  (setq lbs (concat "\n" (make-string (current-column) ?\ ))
1471               ;;        separator nil)
1472               ;;  (while (consp value)
1473               ;;    (setq cell (car value))
1474               ;;    (if (and (consp cell)
1475               ;;             (consp (car cell)))
1476               ;;        (progn
1477               ;;          (if separator
1478               ;;              (insert lbs))
1479               ;;          (char-db-insert-alist cell readable)
1480               ;;          (setq separator lbs))
1481               ;;      (if separator
1482               ;;          (insert separator))
1483               ;;      (insert (prin1-to-string cell))
1484               ;;      (setq separator " "))
1485               ;;    (setq value (cdr value)))
1486               ;;  (insert ")")
1487               ;;  (insert line-breaking))
1488               ((consp value)
1489                (insert (format "(%-18s " name))
1490                (setq lbs (concat "\n" (make-string (current-column) ?\ ))
1491                      separator nil)
1492                (while (consp value)
1493                  (setq cell (car value))
1494                  (if (and (consp cell)
1495                           (consp (car cell))
1496                           (setq ret (condition-case nil
1497                                         (find-char cell)
1498                                       (error nil))))
1499                      (progn
1500                        (setq rest cell
1501                              al nil
1502                              cal nil)
1503                        (while rest
1504                          (setq key (car (car rest)))
1505                          (if (find-charset key)
1506                              (setq cal (cons key cal))
1507                            (setq al (cons key al)))
1508                          (setq rest (cdr rest)))
1509                        (if separator
1510                            (insert lbs))
1511                        (insert-char-attributes ret
1512                                                readable
1513                                                al ; cal
1514                                                nil 'for-sub-node)
1515                        (setq separator lbs))
1516                    (setq ret (prin1-to-string cell))
1517                    (if separator
1518                        (if (< (+ (current-column)
1519                                  (length ret)
1520                                  (length separator))
1521                               76)
1522                            (insert separator)
1523                          (insert lbs)))
1524                    (insert ret)
1525                    (setq separator " "))
1526                  (setq value (cdr value)))
1527                (insert ")")
1528                (insert line-breaking))
1529               (t
1530                (insert (format "(%-18s" name))
1531                (setq ret (prin1-to-string value))
1532                (unless (< (+ (current-column)
1533                              (length ret)
1534                              3)
1535                           76)
1536                  (insert line-breaking))
1537                (insert " . " ret ")" line-breaking)
1538                ;; (insert (format "(%-18s . %S)%s"
1539                ;;                 name value
1540                ;;                 line-breaking))
1541                )
1542               ))
1543       (setq attributes (cdr attributes)))
1544     (insert ")")))
1545
1546 (defun insert-char-data (char &optional readable
1547                               attributes)
1548   (save-restriction
1549     (narrow-to-region (point)(point))
1550     (insert "(define-char
1551   '")
1552     (insert-char-attributes char readable attributes)
1553     (insert ")\n")
1554     (goto-char (point-min))
1555     (while (re-search-forward "[ \t]+$" nil t)
1556       (replace-match ""))
1557     ;; from tabify.
1558     (goto-char (point-min))
1559     (while (re-search-forward "[ \t][ \t][ \t]*" nil t)
1560       (let ((column (current-column))
1561             (indent-tabs-mode t))
1562         (delete-region (match-beginning 0) (point))
1563         (indent-to column)))
1564     (goto-char (point-max))
1565     ;; (tabify (point-min)(point-max))
1566     ))
1567
1568 (defun insert-char-data-with-variant (char &optional printable
1569                                            no-ucs-unified
1570                                            script excluded-script)
1571   (insert-char-data char printable)
1572   (let ((variants (char-variants char))
1573         rest
1574         variant vs ret)
1575     (setq variants (sort variants #'<))
1576     (setq rest variants)
1577     (setq variants (cons char variants))
1578     (while rest
1579       (setq variant (car rest))
1580       (unless (get-char-attribute variant '<-subsumptive)
1581         (if (and (or (null script)
1582                      (null (setq vs (get-char-attribute variant 'script)))
1583                      (memq script vs))
1584                  (or (null excluded-script)
1585                      (null (setq vs (get-char-attribute variant 'script)))
1586                      (not (memq excluded-script vs))))
1587             (unless (and no-ucs-unified (get-char-attribute variant '=ucs))
1588               (insert-char-data variant printable)
1589               (if (setq ret (char-variants variant))
1590                   (while ret
1591                     (or (memq (car ret) variants)
1592                         ;; (get-char-attribute (car ret) '<-subsumptive)
1593                         (setq rest (nconc rest (list (car ret)))))
1594                     (setq ret (cdr ret)))))))
1595       (setq rest (cdr rest)))))
1596
1597 (defun insert-char-range-data (min max &optional script excluded-script)
1598   (let ((code min)
1599         char)
1600     (while (<= code max)
1601       (setq char (decode-char '=ucs code))
1602       (if (encode-char char '=ucs 'defined-only)
1603           (insert-char-data-with-variant char nil 'no-ucs-unified
1604                                          script excluded-script))
1605       (setq code (1+ code)))))
1606
1607 (defun write-char-range-data-to-file (min max file
1608                                           &optional script excluded-script)
1609   (let ((coding-system-for-write char-db-file-coding-system))
1610     (with-temp-buffer
1611       (insert (format ";; -*- coding: %s -*-\n"
1612                       char-db-file-coding-system))
1613       (insert-char-range-data min max script excluded-script)
1614       (write-region (point-min)(point-max) file))))
1615
1616 (defvar what-character-original-window-configuration)
1617
1618 ;;;###autoload
1619 (defun what-char-definition (char)
1620   (interactive (list (char-after)))
1621   (let ((buf (get-buffer-create "*Character Description*"))
1622         (the-buf (current-buffer))
1623         (win-conf (current-window-configuration)))
1624     (pop-to-buffer buf)
1625     (make-local-variable 'what-character-original-window-configuration)
1626     (setq what-character-original-window-configuration win-conf)
1627     (setq buffer-read-only nil)
1628     (erase-buffer)
1629     (condition-case err
1630         (progn
1631           (insert-char-data-with-variant char 'printable)
1632           (unless (char-attribute-alist char)
1633             (insert (format ";; = %c\n"
1634                             (let* ((rest (split-char char))
1635                                    (ccs (pop rest))
1636                                    (code (pop rest)))
1637                               (while rest
1638                                 (setq code (logior (lsh code 8)
1639                                                    (pop rest))))
1640                               (decode-char ccs code)))))
1641           ;; (char-db-update-comment)
1642           (set-buffer-modified-p nil)
1643           (view-mode the-buf (lambda (buf)
1644                                (set-window-configuration
1645                                 what-character-original-window-configuration)
1646                                ))
1647           (goto-char (point-min)))
1648       (error (progn
1649                (set-window-configuration
1650                 what-character-original-window-configuration)
1651                (signal (car err) (cdr err)))))))
1652
1653
1654 ;;; @ end
1655 ;;;
1656
1657 (provide 'char-db-util)
1658
1659 ;;; char-db-util.el ends here