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 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-var-001
 412     =>ucs@iwds-1
 413     =>ucs@cognate
 414     =>ucs@component
 415     =>ucs@iwds-1/normalized
 416     =>ucs-itaiji-001@iwds-1
 417     =>ucs-itaiji-002@iwds-1
 418     =>ucs-itaiji-003@iwds-1
 419     =>ucs-itaiji-006@iwds-1
 420     =>big5-cdp@component
 421     =>big5-cdp@cognate
 422     ==ucs@gb
 423     =ucs@gb
 424     ==ucs-var-002
 425     ==ucs-itaiji-002
 426     =ucs@JP/hanazono
 427     =koseki
 428     =shuowen-jiguge
 429     =>shuowen-jiguge
 430     ===shuowen-jiguge5
 431     ===shuowen-jiguge4
 432     ===chise-hdic-ktb-seal))
 433
 434
 435 ;;; @ char-db formatters
 436 ;;;
 437
 438 (defun char-db-make-char-spec (char)
 439   (let (ret char-spec)
 440     (cond ((characterp char)
 441            (cond ((and (setq ret (encode-char char '=ucs 'defined-only))
 442                        (not (and (<= #xE000 ret)(<= ret #xF8FF))))
 443                   (setq char-spec (list (cons '=ucs ret)))
 444                   (cond ((setq ret (get-char-attribute char 'name))
 445                          (setq char-spec (cons (cons 'name ret) char-spec))
 446                          )
 447                         ((setq ret (get-char-attribute char 'name*))
 448                          (setq char-spec (cons (cons 'name* ret) char-spec))
 449                          ))
 450                   )
 451                  ((encode-char char '=mj 'defined-only)
 452                   (setq char-spec nil)
 453                   (dolist (ccs (charset-list))
 454                     (if (and (or (eq ccs '=mj)
 455                                  ;; (eq (charset-property ccs 'iso-ir) 177)
 456                                  (string-match "=ucs@" (symbol-name ccs))
 457                                  )
 458                              (setq ccs (charset-name ccs))
 459                              (null (assq ccs char-spec))
 460                              (setq ret (encode-char char ccs 'defined-only)))
 461                         (setq char-spec (cons (cons ccs ret) char-spec))))
 462                   )
 463                  ((encode-char char '==mj 'defined-only)
 464                   (setq char-spec nil)
 465                   (dolist (ccs (charset-list))
 466                     (if (and (or (eq ccs '==mj)
 467                                  ;; (eq (charset-property ccs 'iso-ir) 177)
 468                                  (string-match "=ucs@" (symbol-name ccs))
 469                                  )
 470                              (setq ccs (charset-name ccs))
 471                              (null (assq ccs char-spec))
 472                              (setq ret (encode-char char ccs 'defined-only)))
 473                         (setq char-spec (cons (cons ccs ret) char-spec))))
 474                   )
 475                  ((encode-char char '=adobe-japan1 'defined-only)
 476                   (setq char-spec nil)
 477                   (dolist (ccs (charset-list))
 478                     (if (and (or (memq ccs
 479                                        '(=adobe-japan1-0
 480                                          =adobe-japan1-1
 481                                          =adobe-japan1-2
 482                                          =adobe-japan1-3
 483                                          =adobe-japan1-4
 484                                          =adobe-japan1-5
 485                                          =adobe-japan1-6
 486                                          ))
 487                                  ;; (eq (charset-property ccs 'iso-ir) 177)
 488                                  (string-match "=ucs@" (symbol-name ccs))
 489                                  )
 490                              (setq ccs (charset-name ccs))
 491                              (null (assq ccs char-spec))
 492                              (setq ret (encode-char char ccs 'defined-only)))
 493                         (setq char-spec (cons (cons ccs ret) char-spec))))
 494                   )
 495                  ((encode-char char '==adobe-japan1 'defined-only)
 496                   (setq char-spec nil)
 497                   (dolist (ccs (charset-list))
 498                     (if (and (or (memq ccs
 499                                        '(==adobe-japan1-0
 500                                          ==adobe-japan1-1
 501                                          ==adobe-japan1-2
 502                                          ==adobe-japan1-3
 503                                          ==adobe-japan1-4
 504                                          ==adobe-japan1-5
 505                                          ==adobe-japan1-6
 506                                          ))
 507                                  ;; (eq (charset-property ccs 'iso-ir) 177)
 508                                  (string-match "=ucs@" (symbol-name ccs))
 509                                  )
 510                              (setq ccs (charset-name ccs))
 511                              (null (assq ccs char-spec))
 512                              (setq ret (encode-char char ccs 'defined-only)))
 513                         (setq char-spec (cons (cons ccs ret) char-spec))))
 514                   )
 515                  ((setq ret
 516                         (catch 'tag
 517                           (let ((rest char-db-coded-charset-priority-list)
 518                                 ccs)
 519                             (while rest
 520                               (setq ccs (charset-name
 521                                          (find-charset (car rest))))
 522                               (if (setq ret
 523                                         (encode-char char ccs
 524                                                      'defined-only))
 525                                   (throw 'tag (cons ccs ret)))
 526                               (setq rest (cdr rest))))))
 527                   (setq char-spec (list ret))
 528                   (dolist (ccs (delq (car ret) (charset-list)))
 529                     (if (and (or (charset-iso-final-char ccs)
 530                                  (memq ccs
 531                                        '(=daikanwa
 532                                          =daikanwa@rev2
 533                                          ;; =gt-k
 534                                          =jis-x0208@1997
 535                                          ))
 536                                  (and (eq (charset-property ccs 'iso-ir) 177)
 537                                       (not (eq (charset-name ccs) '=ucs@big5)))
 538                                  ;; (string-match "=ucs@" (symbol-name ccs))
 539                                  )
 540                              (setq ccs (charset-name ccs))
 541                              (null (assq ccs char-spec))
 542                              (setq ret (encode-char char ccs 'defined-only)))
 543                         (setq char-spec (cons (cons ccs ret) char-spec))))
 544                   (if (null char-spec)
 545                       (setq char-spec (split-char char)))
 546                   (cond ((setq ret (get-char-attribute char 'name))
 547                          (setq char-spec (cons (cons 'name ret) char-spec))
 548                          )
 549                         ((setq ret (get-char-attribute char 'name*))
 550                          (setq char-spec (cons (cons 'name* ret) char-spec))
 551                          ))
 552                   )
 553                  ((setq ret (get-char-attribute
 554                              char 'ideographic-combination))
 555                   (setq char-spec
 556                         (cons (cons 'ideographic-combination ret)
 557                               char-spec))
 558                   ))
 559            char-spec)
 560           ((consp char)
 561            char))))
 562
 563 (defun char-db-insert-char-spec (char &optional readable column
 564                                       required-features)
 565   (unless column
 566     (setq column (current-column)))
 567   (let (char-spec temp-char)
 568     (setq char-spec (char-db-make-char-spec char))
 569     (unless (or (characterp char) ; char
 570                 (condition-case nil
 571                     (setq char (find-char char-spec))
 572                   (error nil)))
 573       ;; define temporary character
 574       ;;   Current implementation is dirty.
 575       (setq temp-char (define-char (cons '(ideograph-daikanwa . 0)
 576                                          char-spec)))
 577       (remove-char-attribute temp-char 'ideograph-daikanwa)
 578       (setq char temp-char))
 579     (insert-char-attributes char
 580                             readable
 581                             (union (mapcar #'car char-spec)
 582                                    required-features)
 583                             nil 'for-sub-node)
 584     (when temp-char
 585       ;; undefine temporary character
 586       ;;   Current implementation is dirty.
 587       (setq char-spec (char-attribute-alist temp-char))
 588       (while char-spec
 589         (remove-char-attribute temp-char (car (car char-spec)))
 590         (setq char-spec (cdr char-spec))))))
 591
 592 (defun char-db-insert-alist (alist &optional readable column)
 593   (unless column
 594     (setq column (current-column)))
 595   (let ((line-breaking
 596          (concat "\n" (make-string (1+ column) ?\ )))
 597         name value
 598         ret al ; cal
 599         key
 600         lbs cell rest separator)
 601     (insert "(")
 602     (while alist
 603       (setq name (car (car alist))
 604             value (cdr (car alist)))
 605       (cond ((eq name 'char)
 606              (insert "(char . ")
 607              (if (setq ret (condition-case nil
 608                                (find-char value)
 609                              (error nil)))
 610                  (progn
 611                    (setq al nil
 612                          ;; cal nil
 613                          )
 614                    (while value
 615                      (setq key (car (car value)))
 616                      ;; (if (find-charset key)
 617                      ;;     (setq cal (cons key cal))
 618                      (setq al (cons key al))
 619                      ;; )
 620                      (setq value (cdr value)))
 621                    (insert-char-attributes ret
 622                                            readable
 623                                            (or al 'none) ; cal
 624                                            nil 'for-sub-node))
 625                (insert (prin1-to-string value)))
 626              (insert ")")
 627              (insert line-breaking))
 628             ((consp value)
 629              (insert (format "(%-18s " name))
 630              (setq lbs (concat "\n" (make-string (current-column) ?\ )))
 631              (while (consp value)
 632                (setq cell (car value))
 633                (if (and (consp cell)
 634                         (consp (car cell))
 635                         (setq ret (condition-case nil
 636                                       (find-char cell)
 637                                     (error nil)))
 638                         )
 639                    (progn
 640                      (setq rest cell
 641                            al nil
 642                            ;; cal nil
 643                            )
 644                      (while rest
 645                        (setq key (car (car rest)))
 646                        ;; (if (find-charset key)
 647                        ;;     (setq cal (cons key cal))
 648                        (setq al (cons key al))
 649                        ;; )
 650                        (setq rest (cdr rest)))
 651                      (if separator
 652                          (insert lbs))
 653                      (insert-char-attributes ret
 654                                              readable
 655                                              al ; cal
 656                                              nil 'for-sub-node)
 657                      (setq separator lbs))
 658                  (if separator
 659                      (insert separator))
 660                  (insert (prin1-to-string cell))
 661                  (setq separator " "))
 662                (setq value (cdr value)))
 663              (insert ")")
 664              (insert line-breaking))
 665             (t
 666              (insert (format "(%-18s . %S)%s"
 667                              name value
 668                              line-breaking))))
 669       (setq alist (cdr alist))))
 670   (insert ")"))
 671
 672 (defun char-db-insert-char-reference (plist &optional readable column)
 673   (unless column
 674     (setq column (current-column)))
 675   (let ((line-breaking
 676          (concat "\n" (make-string (1+ column) ?\ )))
 677         (separator "")
 678         name value)
 679     (insert "(")
 680     (while plist
 681       (setq name (pop plist))
 682       (setq value (pop plist))
 683       (cond ((eq name :char)
 684              (insert separator)
 685              (insert ":char\t")
 686              (cond ((numberp value)
 687                     (setq value (decode-char '=ucs value)))
 688                    ;; ((consp value)
 689                    ;;  (setq value (or (find-char value)
 690                    ;;                  value)))
 691                    )
 692              (char-db-insert-char-spec value readable)
 693              (insert line-breaking)
 694              (setq separator ""))
 695             ((eq name :radical)
 696              (insert (format "%s%s\t%d ; %c%s"
 697                              separator
 698                              name value
 699                              (ideographic-radical value)
 700                              line-breaking))
 701              (setq separator ""))
 702             (t
 703              (insert (format "%s%s\t%S" separator name value))
 704              (setq separator line-breaking)))
 705       ))
 706   (insert ")"))
 707
 708 (defun char-db-decode-isolated-char (ccs code-point)
 709   (let (ret)
 710     (setq ret
 711           (cond ((eq ccs 'arabic-iso8859-6)
 712                  (decode-char ccs code-point))
 713                 ;; ((eq ccs '=gt)
 714                 ;;  (decode-builtin-char '==gt code-point))
 715                 ((and (memq ccs '(=gt-pj-1
 716                                   =gt-pj-2
 717                                   =gt-pj-3
 718                                   =gt-pj-4
 719                                   =gt-pj-5
 720                                   =gt-pj-6
 721                                   =gt-pj-7
 722                                   =gt-pj-8
 723                                   =gt-pj-9
 724                                   =gt-pj-10
 725                                   =gt-pj-11))
 726                       (setq ret (decode-char ccs code-point))
 727                       (setq ret (encode-char ret '=gt 'defined-only)))
 728                  (decode-builtin-char '=gt ret))
 729                 (t
 730                  (decode-builtin-char ccs code-point))))
 731     (cond ((null ret)
 732            (or (decode-char ccs code-point)
 733                (define-char (list (cons ccs code-point)))))
 734           ((and (<= 0 (char-int ret))
 735                 (<= (char-int ret) #x1F))
 736            (decode-char '=ucs (+ #x2400 (char-int ret))))
 737           ((= (char-int ret) #x7F)
 738            ?\u2421)
 739           (t ret))))
 740
 741 (defvar char-db-convert-obsolete-format t)
 742
 743 (defun char-db-insert-ccs-feature (name value line-breaking)
 744   (cond
 745    ((integerp value)
 746     (insert
 747      (format
 748       (cond
 749        ((memq name '(=>iwds-1
 750                      ==shinjigen
 751                      =shinjigen
 752                      =shinjigen@1ed ==shinjigen@1ed
 753                      =shinjigen@rev ==shinjigen@rev
 754                      =shinjigen/+p@rev ==shinjigen/+p@rev
 755                      ===daikanwa/ho ==daikanwa/ho
 756                      =daikanwa/ho =>>daikanwa/ho =>daikanwa/ho))
 757         "(%-18s .  %04d)\t; %c")
 758        ((eq name '=shinjigen@1ed/24pr)
 759         "(%-18s . %04d)\t; %c")
 760        ((or
 761          (memq name
 762                '(===daikanwa
 763                  ==daikanwa =daikanwa =>>daikanwa =>daikanwa
 764                  =daikanwa@rev1 =daikanwa@rev2
 765                  =daikanwa/+p ==daikanwa/+p ===daikanwa/+p
 766                  =>>daikanwa/+p
 767                  =daikanwa/+2p =>>daikanwa/+2p
 768                  =gt ==gt ===gt
 769                  =>>gt =+>gt =>gt
 770                  =gt-k ==gt-k ===gt-k
 771                  =>>gt-k =>gt-k
 772                  =adobe-japan1-0 ==adobe-japan1-0 ===adobe-japan1-0
 773                  =adobe-japan1-1 ==adobe-japan1-1 ===adobe-japan1-1
 774                  =adobe-japan1-2 ==adobe-japan1-2 ===adobe-japan1-2
 775                  =adobe-japan1-3 ==adobe-japan1-3 ===adobe-japan1-3
 776                  =adobe-japan1-4 ==adobe-japan1-4 ===adobe-japan1-4
 777                  =adobe-japan1-5 ==adobe-japan1-5 ===adobe-japan1-5
 778                  =adobe-japan1-6 ==adobe-japan1-6 ===adobe-japan1-6
 779                  =>>adobe-japan1-0 =+>adobe-japan1-0
 780                  =>>adobe-japan1-1 =+>adobe-japan1-1
 781                  =>>adobe-japan1-2 =+>adobe-japan1-2
 782                  =>>adobe-japan1-3 =+>adobe-japan1-3
 783                  =>>adobe-japan1-4 =+>adobe-japan1-4
 784                  =>>adobe-japan1-5 =+>adobe-japan1-5
 785                  =>>adobe-japan1-6 =+>adobe-japan1-6
 786                  =>cbeta =cbeta =>>cbeta ==cbeta ===cbeta
 787                  =zinbun-oracle =>zinbun-oracle
 788                  =daijiten ==daijiten ===daijiten ==>daijiten
 789                  ===hng-jou ===hng-keg ===hng-dng ===hng-mam
 790                  ===hng-drt ===hng-kgk ===hng-myz ===hng-kda
 791                  ===hng-khi ===hng-khm ===hng-fhs ===hng-hok
 792                  ===hng-kyd ===hng-sok
 793                  ===hng-yhk ===hng-kak ===hng-kar ===hng-kae
 794                  ===hng-sys ===hng-tsu ===hng-tzj
 795                  ===hng-hos ===hng-kkh ===hng-nak ===hng-jhk
 796                  ===hng-hod ===hng-gok ===hng-ink ===hng-nto
 797                  ===hng-nkm ===hng-k24 ===hng-ini ===hng-nkk
 798                  ===hng-kcc ===hng-kcj ===hng-kbk ===hng-sik
 799                  ===hng-skk ===hng-kyu ===hng-ksk ===hng-wan
 800                  ===hng-okd ===hng-wad ===hng-kmi
 801                  ===hng-zkd ===hng-doh ===hng-jyu ===hng-tzs
 802                  ===hng-sai ===hng-kad ===hng-kss
 803                  ===hng-kyo ===hng-ykk ===hng-saa
 804                  ===hng-sab ===hng-wks ===hng-wke ===hng-smk
 805                  ===hng-sgs ===hng-sts ===hng-kkd
 806                  =>shuowen-jiguge =shuowen-jiguge
 807                  ===shuowen-jiguge4 ===shuowen-jiguge5 ===shuowen-jiguge-A30))
 808          ;; (string-match "^=adobe-" (symbol-name name))
 809          )
 810         "(%-18s . %05d)\t; %c")
 811        ((memq name '(=hanyo-denshi/ks
 812                      ==hanyo-denshi/ks ===hanyo-denshi/ks
 813                      =>>hanyo-denshi/ks
 814                      =koseki ==koseki
 815                      =mj ==mj ===mj =>>mj =>mj
 816                      =zihai mojikyo))
 817         "(%-18s . %06d)\t; %c")
 818        ((memq name '(=hanyo-denshi/tk =>hanyo-denshi/tk ==hanyo-denshi/tk))
 819         "(%-18s . %08d)\t; %c")
 820        ((>= (charset-dimension name) 2)
 821         "(%-18s . #x%04X)\t; %c")
 822        (t
 823         "(%-18s . #x%02X)\t; %c"))
 824       name
 825       (if (= (charset-iso-graphic-plane name) 1)
 826           (logior value
 827                   (cond ((= (charset-dimension name) 1)
 828                          #x80)
 829                         ((= (charset-dimension name) 2)
 830                          #x8080)
 831                         ((= (charset-dimension name) 3)
 832                          #x808080)
 833                         (t 0)))
 834         value)
 835       (char-db-decode-isolated-char name value)))
 836     (if (and (= (charset-chars name) 94)
 837              (= (charset-dimension name) 2))
 838         (insert (format " [%02d-%02d]"
 839                         (- (lsh value -8) 32)
 840                         (- (logand value 255) 32))))
 841     )
 842    (t
 843     (insert (format "(%-18s . %s)" name value))
 844     ))
 845   (insert line-breaking))
 846
 847 (defun char-db-insert-relation-feature (char name value line-breaking
 848                                              ccss readable)
 849   (insert (format "(%-18s%s " name line-breaking))
 850   (let ((lbs (concat "\n" (make-string (current-column) ?\ )))
 851         separator cell sources required-features
 852         ret)
 853     (while (consp value)
 854       (setq cell (car value))
 855       (if (integerp cell)
 856           (setq cell (decode-char '=ucs cell)))
 857       (cond
 858        ((eq name '->subsumptive)
 859         (when (or (not (some (lambda (atr)
 860                                (get-char-attribute cell atr))
 861                              char-db-ignored-attributes))
 862                   (some (lambda (ccs)
 863                           (encode-char cell ccs 'defined-only))
 864                         ccss))
 865           (if separator
 866               (insert lbs))
 867           (let ((char-db-ignored-attributes
 868                  (cons '<-subsumptive
 869                        char-db-ignored-attributes)))
 870             (insert-char-attributes cell readable nil nil 'for-sub-node))
 871           (setq separator lbs))
 872         )
 873        ((characterp cell)
 874         (setq sources
 875               (get-char-attribute
 876                char (intern (format "%s*sources" name))))
 877         (setq required-features nil)
 878         (dolist (source sources)
 879           (cond
 880            ((memq source '(JP
 881                            JP/Jouyou
 882                            shinjigen shinjigen@1ed shinjigen@rev))
 883             (setq required-features
 884                   (union required-features
 885                          '(=jis-x0208
 886                            =jis-x0208@1990
 887                            =jis-x0213-1@2000
 888                            =jis-x0213-1@2004
 889                            =jis-x0213-2
 890                            =jis-x0212
 891                            =jis-x0208@1983
 892                            =jis-x0208@1978
 893                            =shinjigen
 894                            =shinjigen@1ed
 895                            =shinjigen@rev
 896                            =shinjigen/+p@rev))))
 897            ((eq source 'CN)
 898             (setq required-features
 899                   (union required-features
 900                          '(=gb2312
 901                            =gb12345
 902                            =iso-ir165)))))
 903           (cond
 904            ((find-charset (setq ret (intern (format "=%s" source))))
 905             (setq required-features
 906                   (cons ret required-features)))
 907            (t (setq required-features
 908                     (cons source required-features)))))
 909         (cond ((string-match "@JP" (symbol-name name))
 910                (setq required-features
 911                      (union required-features
 912                             '(=jis-x0208
 913                               =jis-x0208@1990
 914                               =jis-x0213-1-2000
 915                               =jis-x0213-2-2000
 916                               =jis-x0212
 917                               =jis-x0208@1983
 918                               =jis-x0208@1978))))
 919               ((string-match "@CN" (symbol-name name))
 920                (setq required-features
 921                      (union required-features
 922                             '(=gb2312
 923                               =gb12345
 924                               =iso-ir165)))))
 925         (if separator
 926             (insert lbs))
 927         (if readable
 928             (insert (format "%S" cell))
 929           (char-db-insert-char-spec cell readable
 930                                     nil
 931                                     required-features))
 932         (setq separator lbs))
 933        ((consp cell)
 934         (if separator
 935             (insert lbs))
 936         (if (consp (car cell))
 937             (char-db-insert-char-spec cell readable)
 938           (char-db-insert-char-reference cell readable))
 939         (setq separator lbs))
 940        (t
 941         (if separator
 942             (insert separator))
 943         (insert (prin1-to-string cell))
 944         (setq separator " ")))
 945       (setq value (cdr value)))
 946     (insert ")")
 947     (insert line-breaking)))
 948
 949 (defun insert-char-attributes (char &optional readable attributes column
 950                                     for-sub-node)
 951   (unless column
 952     (setq column (current-column)))
 953   (let (name value ; has-long-ccs-name
 954         rest
 955         radical strokes
 956         (line-breaking
 957          (concat "\n" (make-string (1+ column) ?\ )))
 958         lbs cell separator ret
 959         key al cal
 960         dest-ccss ; sources required-features
 961         ccss)
 962     (let (atr-d)
 963       (setq attributes
 964             (sort (if attributes
 965                       (if (consp attributes)
 966                           (progn
 967                             (dolist (name attributes)
 968                               (unless (memq name char-db-ignored-attributes)
 969                                 (if (find-charset name)
 970                                     (push name ccss))
 971                                 (push name atr-d)))
 972                             atr-d))
 973                     (dolist (name (char-attribute-list))
 974                       (unless (memq name char-db-ignored-attributes)
 975                         (if (find-charset name)
 976                             (push name ccss))
 977                         (push name atr-d)))
 978                     atr-d)
 979                   #'char-attribute-name<)))
 980     (insert "(")
 981     (when (memq '<-subsumptive attributes)
 982       (when (or readable (not for-sub-node))
 983         (when (setq value (get-char-attribute char '<-subsumptive))
 984           (char-db-insert-relation-feature char '<-subsumptive value
 985                                            line-breaking
 986                                            ccss readable)))
 987       (setq attributes (delq '<-subsumptive attributes)))
 988     (when (and (memq '<-denotational attributes)
 989                (setq value (get-char-attribute char '<-denotational)))
 990       (char-db-insert-relation-feature char '<-denotational value
 991                                        line-breaking
 992                                        ccss readable)
 993       (setq attributes (delq '<-denotational attributes)))
 994     (when (and (memq '<-denotational@component attributes)
 995                (setq value (get-char-attribute char '<-denotational@component)))
 996       (char-db-insert-relation-feature char '<-denotational@component value
 997                                        line-breaking
 998                                        ccss readable)
 999       (setq attributes (delq '<-denotational@component attributes)))
1000     (when (and (memq '<-denotational@usage attributes)
1001                (setq value (get-char-attribute char '<-denotational@usage)))
1002       (char-db-insert-relation-feature char '<-denotational@usage value
1003                                        line-breaking
1004                                        ccss readable)
1005       (setq attributes (delq '<-denotational@usage attributes)))
1006     (when (and (memq 'name attributes)
1007                (setq value (get-char-attribute char 'name)))
1008       (insert (format
1009                (if (> (+ (current-column) (length value)) 48)
1010                    "(name . %S)%s"
1011                  "(name               . %S)%s")
1012                value line-breaking))
1013       (setq attributes (delq 'name attributes))
1014       )
1015     (when (and (memq 'name* attributes)
1016                (setq value (get-char-attribute char 'name*)))
1017       (insert (format
1018                (if (> (+ (current-column) (length value)) 48)
1019                    "(name* . %S)%s"
1020                  "(name*              . %S)%s")
1021                value line-breaking))
1022       (setq attributes (delq 'name* attributes))
1023       )
1024     (when (and (memq 'script attributes)
1025                (setq value (get-char-attribute char 'script)))
1026       (insert (format "(script\t\t%s)%s"
1027                       (mapconcat (function prin1-to-string)
1028                                  value " ")
1029                       line-breaking))
1030       (setq attributes (delq 'script attributes))
1031       )
1032     (dolist (name '(=>ucs =>ucs*))
1033       (when (and (memq name attributes)
1034                  (setq value (get-char-attribute char name)))
1035         (insert (format "(%-18s . #x%04X)\t; %c%s"
1036                         name value (decode-char '=ucs value)
1037                         line-breaking))
1038         (setq attributes (delq name attributes))))
1039     (dolist (name '(=>ucs@gb =>ucs@big5))
1040       (when (and (memq name attributes)
1041                  (setq value (get-char-attribute char name)))
1042         (insert (format "(%-18s . #x%04X)\t; %c%s"
1043                         name value
1044                         (decode-char (intern
1045                                       (concat "="
1046                                               (substring
1047                                                (symbol-name name) 2)))
1048                                      value)
1049                         line-breaking))
1050         (setq attributes (delq name attributes))
1051         ))
1052     ;; (dolist (name '(=>daikanwa))
1053     ;;   (when (and (memq name attributes)
1054     ;;              (setq value (get-char-attribute char name)))
1055     ;;     (insert
1056     ;;      (if (integerp value)
1057     ;;          (format "(%-18s . %05d)\t; %c%s"
1058     ;;                  name value (decode-char '=daikanwa value)
1059     ;;                  line-breaking)
1060     ;;        (format "(%-18s %s)\t; %c%s"
1061     ;;                name
1062     ;;                (mapconcat (function prin1-to-string)
1063     ;;                           value " ")
1064     ;;                (char-representative-of-daikanwa char)
1065     ;;                line-breaking)))
1066     ;;     (setq attributes (delq name attributes))))
1067     (when (and (memq 'general-category attributes)
1068                (setq value (get-char-attribute char 'general-category)))
1069       (insert (format
1070                "(general-category\t%s) ; %s%s"
1071                (mapconcat (lambda (cell)
1072                             (format "%S" cell))
1073                           value " ")
1074                (cond ((rassoc value unidata-normative-category-alist)
1075                       "Normative Category")
1076                      ((rassoc value unidata-informative-category-alist)
1077                       "Informative Category")
1078                      (t
1079                       "Unknown Category"))
1080                line-breaking))
1081       (setq attributes (delq 'general-category attributes))
1082       )
1083     (when (and (memq 'bidi-category attributes)
1084                (setq value (get-char-attribute char 'bidi-category)))
1085       (insert (format "(bidi-category\t. %S)%s"
1086                       value
1087                       line-breaking))
1088       (setq attributes (delq 'bidi-category attributes))
1089       )
1090     (unless (or (not (memq 'mirrored attributes))
1091                 (eq (setq value (get-char-attribute char 'mirrored 'empty))
1092                     'empty))
1093       (insert (format "(mirrored\t\t. %S)%s"
1094                       value
1095                       line-breaking))
1096       (setq attributes (delq 'mirrored attributes))
1097       )
1098     (cond
1099      ((and (memq 'decimal-digit-value attributes)
1100            (setq value (get-char-attribute char 'decimal-digit-value)))
1101       (insert (format "(decimal-digit-value . %S)%s"
1102                       value
1103                       line-breaking))
1104       (setq attributes (delq 'decimal-digit-value attributes))
1105       (when (and (memq 'digit-value attributes)
1106                  (setq value (get-char-attribute char 'digit-value)))
1107         (insert (format "(digit-value\t . %S)%s"
1108                         value
1109                         line-breaking))
1110         (setq attributes (delq 'digit-value attributes))
1111         )
1112       (when (and (memq 'numeric-value attributes)
1113                  (setq value (get-char-attribute char 'numeric-value)))
1114         (insert (format "(numeric-value\t . %S)%s"
1115                         value
1116                         line-breaking))
1117         (setq attributes (delq 'numeric-value attributes))
1118         )
1119       )
1120      (t
1121       (when (and (memq 'digit-value attributes)
1122                  (setq value (get-char-attribute char 'digit-value)))
1123         (insert (format "(digit-value\t. %S)%s"
1124                         value
1125                         line-breaking))
1126         (setq attributes (delq 'digit-value attributes))
1127         )
1128       (when (and (memq 'numeric-value attributes)
1129                  (setq value (get-char-attribute char 'numeric-value)))
1130         (insert (format "(numeric-value\t. %S)%s"
1131                         value
1132                         line-breaking))
1133         (setq attributes (delq 'numeric-value attributes))
1134         )))
1135     (when (and (memq 'iso-10646-comment attributes)
1136                (setq value (get-char-attribute char 'iso-10646-comment)))
1137       (insert (format "(iso-10646-comment\t. %S)%s"
1138                       value
1139                       line-breaking))
1140       (setq attributes (delq 'iso-10646-comment attributes))
1141       )
1142     (when (and (memq 'morohashi-daikanwa attributes)
1143                (setq value (get-char-attribute char 'morohashi-daikanwa)))
1144       (insert (format "(morohashi-daikanwa\t%s)%s"
1145                       (mapconcat (function prin1-to-string) value " ")
1146                       line-breaking))
1147       (setq attributes (delq 'morohashi-daikanwa attributes))
1148       )
1149     (setq radical nil
1150           strokes nil)
1151     (when (and (memq 'ideographic-radical attributes)
1152                (setq value (get-char-attribute char 'ideographic-radical)))
1153       (setq radical value)
1154       (insert (format "(ideographic-radical . %S)\t; %c%s"
1155                       radical
1156                       (ideographic-radical radical)
1157                       line-breaking))
1158       (setq attributes (delq 'ideographic-radical attributes))
1159       )
1160     (when (and (memq 'shuowen-radical attributes)
1161                (setq value (get-char-attribute char 'shuowen-radical)))
1162       (insert (format "(shuowen-radical\t. %S)\t; %c%s"
1163                       value
1164                       (shuowen-radical value)
1165                       line-breaking))
1166       (setq attributes (delq 'shuowen-radical attributes))
1167       )
1168     (let (key)
1169       (dolist (domain
1170                (append
1171                 char-db-feature-domains
1172                 (let (dest domain)
1173                   (dolist (feature (char-attribute-list))
1174                     (setq feature (symbol-name feature))
1175                     (when (string-match
1176                            "\\(radical\\|strokes\\)@\\([^@*]+\\)\\(\\*\\|$\\)"
1177                            feature)
1178                       (setq domain (intern (match-string 2 feature)))
1179                      (unless (memq domain dest)
1180                        (setq dest (cons domain dest)))))
1181                   (sort dest #'string<))))
1182         (setq key (intern (format "%s@%s" 'ideographic-radical domain)))
1183         (when (and (memq key attributes)
1184                    (setq value (get-char-attribute char key)))
1185           (setq radical value)
1186           (insert (format "(%s . %S)\t; %c%s"
1187                           key
1188                           radical
1189                           (ideographic-radical radical)
1190                           line-breaking))
1191           (setq attributes (delq key attributes))
1192           )
1193         (setq key (intern (format "%s@%s" 'ideographic-strokes domain)))
1194         (when (and (memq key attributes)
1195                    (setq value (get-char-attribute char key)))
1196           (setq strokes value)
1197           (insert (format "(%s . %S)%s"
1198                           key
1199                           strokes
1200                           line-breaking))
1201           (setq attributes (delq key attributes))
1202           )
1203         (setq key (intern (format "%s@%s" 'total-strokes domain)))
1204         (when (and (memq key attributes)
1205                    (setq value (get-char-attribute char key)))
1206           (insert (format "(%s       . %S)%s"
1207                           key
1208                           value
1209                           line-breaking))
1210           (setq attributes (delq key attributes))
1211           )
1212         (dolist (feature '(ideographic-radical
1213                            ideographic-strokes
1214                            total-strokes))
1215           (setq key (intern (format "%s@%s*sources" feature domain)))
1216           (when (and (memq key attributes)
1217                      (setq value (get-char-attribute char key)))
1218             (insert (format "(%s%s" key line-breaking))
1219             (dolist (cell value)
1220               (insert (format " %s" cell)))
1221             (insert ")")
1222             (insert line-breaking)
1223             (setq attributes (delq key attributes))
1224             ))
1225         ))
1226     (when (and (memq 'ideographic-strokes attributes)
1227                (setq value (get-char-attribute char 'ideographic-strokes)))
1228       (setq strokes value)
1229       (insert (format "(ideographic-strokes . %S)%s"
1230                       strokes
1231                       line-breaking))
1232       (setq attributes (delq 'ideographic-strokes attributes))
1233       )
1234     (when (and (memq 'kangxi-radical attributes)
1235                (setq value (get-char-attribute char 'kangxi-radical)))
1236       (unless (eq value radical)
1237         (insert (format "(kangxi-radical\t . %S)\t; %c%s"
1238                         value
1239                         (ideographic-radical value)
1240                         line-breaking))
1241         (or radical
1242             (setq radical value)))
1243       (setq attributes (delq 'kangxi-radical attributes))
1244       )
1245     (when (and (memq 'kangxi-strokes attributes)
1246                (setq value (get-char-attribute char 'kangxi-strokes)))
1247       (unless (eq value strokes)
1248         (insert (format "(kangxi-strokes\t . %S)%s"
1249                         value
1250                         line-breaking))
1251         (or strokes
1252             (setq strokes value)))
1253       (setq attributes (delq 'kangxi-strokes attributes))
1254       )
1255     ;; (when (and (memq 'japanese-radical attributes)
1256     ;;            (setq value (get-char-attribute char 'japanese-radical)))
1257     ;;   (unless (eq value radical)
1258     ;;     (insert (format "(japanese-radical\t . %S)\t; %c%s"
1259     ;;                     value
1260     ;;                     (ideographic-radical value)
1261     ;;                     line-breaking))
1262     ;;     (or radical
1263     ;;         (setq radical value)))
1264     ;;   (setq attributes (delq 'japanese-radical attributes))
1265     ;;   )
1266     (when (and (memq 'japanese-strokes attributes)
1267                (setq value (get-char-attribute char 'japanese-strokes)))
1268       (unless (eq value strokes)
1269         (insert (format "(japanese-strokes\t . %S)%s"
1270                         value
1271                         line-breaking))
1272         (or strokes
1273             (setq strokes value)))
1274       (setq attributes (delq 'japanese-strokes attributes))
1275       )
1276     (when (and (memq 'cns-radical attributes)
1277                (setq value (get-char-attribute char 'cns-radical)))
1278       (insert (format "(cns-radical\t . %S)\t; %c%s"
1279                       value
1280                       (ideographic-radical value)
1281                       line-breaking))
1282       (setq attributes (delq 'cns-radical attributes))
1283       )
1284     (when (and (memq 'cns-strokes attributes)
1285                (setq value (get-char-attribute char 'cns-strokes)))
1286       (unless (eq value strokes)
1287         (insert (format "(cns-strokes\t . %S)%s"
1288                         value
1289                         line-breaking))
1290         (or strokes
1291             (setq strokes value)))
1292       (setq attributes (delq 'cns-strokes attributes))
1293       )
1294     ;; (when (and (memq 'ideographic- attributes)
1295     ;;            (setq value (get-char-attribute char 'ideographic-)))
1296     ;;   (insert "(ideographic-       ")
1297     ;;   (setq lbs (concat "\n" (make-string (current-column) ?\ ))
1298     ;;         separator nil)
1299     ;;   (while (consp value)
1300     ;;     (setq cell (car value))
1301     ;;     (if (integerp cell)
1302     ;;         (setq cell (decode-char '=ucs cell)))
1303     ;;     (cond ((characterp cell)
1304     ;;            (if separator
1305     ;;                (insert lbs))
1306     ;;            (if readable
1307     ;;                (insert (format "%S" cell))
1308     ;;              (char-db-insert-char-spec cell readable))
1309     ;;            (setq separator lbs))
1310     ;;           ((consp cell)
1311     ;;            (if separator
1312     ;;                (insert lbs))
1313     ;;            (if (consp (car cell))
1314     ;;                (char-db-insert-char-spec cell readable)
1315     ;;              (char-db-insert-char-reference cell readable))
1316     ;;            (setq separator lbs))
1317     ;;           (t
1318     ;;            (if separator
1319     ;;                (insert separator))
1320     ;;            (insert (prin1-to-string cell))
1321     ;;            (setq separator " ")))
1322     ;;     (setq value (cdr value)))
1323     ;;   (insert ")")
1324     ;;   (insert line-breaking)
1325     ;;   (setq attributes (delq 'ideographic- attributes)))
1326     (when (and (memq 'total-strokes attributes)
1327                (setq value (get-char-attribute char 'total-strokes)))
1328       (insert (format "(total-strokes       . %S)%s"
1329                       value
1330                       line-breaking))
1331       (setq attributes (delq 'total-strokes attributes))
1332       )
1333     ;; (when (and (memq '->ideograph attributes)
1334     ;;            (setq value (get-char-attribute char '->ideograph)))
1335     ;;   (insert (format "(->ideograph\t%s)%s"
1336     ;;                   (mapconcat (lambda (code)
1337     ;;                                (cond ((symbolp code)
1338     ;;                                       (symbol-name code))
1339     ;;                                      ((integerp code)
1340     ;;                                       (format "#x%04X" code))
1341     ;;                                      (t
1342     ;;                                       (format "%s %S"
1343     ;;                                               line-breaking code))))
1344     ;;                              value " ")
1345     ;;                   line-breaking))
1346     ;;   (setq attributes (delq '->ideograph attributes))
1347     ;;   )
1348     ;; (when (and (memq '->decomposition attributes)
1349     ;;            (setq value (get-char-attribute char '->decomposition)))
1350     ;;   (insert (format "(->decomposition\t%s)%s"
1351     ;;                   (mapconcat (lambda (code)
1352     ;;                                (cond ((symbolp code)
1353     ;;                                       (symbol-name code))
1354     ;;                                      ((characterp code)
1355     ;;                                       (if readable
1356     ;;                                           (format "%S" code)
1357     ;;                                         (format "#x%04X"
1358     ;;                                                 (char-int code))
1359     ;;                                         ))
1360     ;;                                      ((integerp code)
1361     ;;                                       (format "#x%04X" code))
1362     ;;                                      (t
1363     ;;                                       (format "%s%S" line-breaking code))))
1364     ;;                              value " ")
1365     ;;                   line-breaking))
1366     ;;   (setq attributes (delq '->decomposition attributes))
1367     ;;   )
1368     (if (equal (get-char-attribute char '->titlecase)
1369                (get-char-attribute char '->uppercase))
1370         (setq attributes (delq '->titlecase attributes)))
1371     (when (and (memq '->mojikyo attributes)
1372                (setq value (get-char-attribute char '->mojikyo)))
1373       (insert (format "(->mojikyo\t\t. %06d)\t; %c%s"
1374                       value (decode-char 'mojikyo value)
1375                       line-breaking))
1376       (setq attributes (delq '->mojikyo attributes))
1377       )
1378     (unless readable
1379       (dolist (ignored '(composition
1380                          ->denotational <-subsumptive ->ucs-unified
1381                          ;; ->ideographic-component-forms
1382                          ))
1383         (setq attributes (delq ignored attributes))))
1384     (while attributes
1385       (setq name (car attributes))
1386       (unless (eq (setq value (get-char-attribute char name 'value-is-empty))
1387                   'value-is-empty)
1388         (cond ((setq ret (find-charset name))
1389                (setq name (charset-name ret))
1390                (when (not (memq name dest-ccss))
1391                  (setq dest-ccss (cons name dest-ccss))
1392                  (char-db-insert-ccs-feature name value line-breaking))
1393                )
1394               ((string-match "^=>ucs@" (symbol-name name))
1395                (insert (format "(%-18s . #x%04X)\t; %c%s"
1396                                name value (decode-char '=ucs value)
1397                                line-breaking))
1398                )
1399               ((eq name 'jisx0208-1978/4X)
1400                (insert (format "(%-18s . #x%04X)%s"
1401                                name value
1402                                line-breaking))
1403                )
1404               ((and
1405                 (not readable)
1406                 (not (eq name '->subsumptive))
1407                 (not (eq name '->uppercase))
1408                 (not (eq name '->lowercase))
1409                 (not (eq name '->titlecase))
1410                 (not (eq name '->canonical))
1411                 (not (eq name '->Bopomofo))
1412                 (not (eq name '->mistakable))
1413                 (not (eq name '->ideographic-variants))
1414                 (null (get-char-attribute
1415                        char (intern (format "%s*sources" name))))
1416                 (not (string-match "\\*sources$" (symbol-name name)))
1417                 (null (get-char-attribute
1418                        char (intern (format "%s*note" name))))
1419                 (not (string-match "\\*note$" (symbol-name name)))
1420                 (or (eq name '<-identical)
1421                     (eq name '<-uppercase)
1422                     (eq name '<-lowercase)
1423                     (eq name '<-titlecase)
1424                     (eq name '<-canonical)
1425                     (eq name '<-ideographic-variants)
1426                     ;; (eq name '<-synonyms)
1427                     (string-match "^<-synonyms" (symbol-name name))
1428                     (eq name '<-mistakable)
1429                     (when (string-match "^->" (symbol-name name))
1430                       (cond
1431                        ((string-match "^->fullwidth" (symbol-name name))
1432                         (not (and (consp value)
1433                                   (characterp (car value))
1434                                   (encode-char
1435                                    (car value) '=ucs 'defined-only)))
1436                         )
1437                        (t)))
1438                     ))
1439                )
1440               ((or (eq name 'ideographic-structure)
1441                    (eq name 'ideographic-structure@apparent)
1442                    (eq name 'ideographic-structure@apparent/leftmost)
1443                    (eq name 'ideographic-structure@apparent/rightmost)
1444                    (eq name 'ideographic-combination)
1445                    ;; (eq name 'ideographic-)
1446                    (eq name '=decomposition)
1447                    (char-feature-base-name= '=decomposition name)
1448                    (char-feature-base-name= '=>decomposition name)
1449                    ;; (string-match "^=>*decomposition\\(@[^*]+\\)?$"
1450                    ;;               (symbol-name name))
1451                    (string-match "^\\(->\\|<-\\)[^*]*$" (symbol-name name))
1452                    (string-match "^\\(->\\|<-\\)[^*]*\\*sources$"
1453                                  (symbol-name name))
1454                    )
1455                (char-db-insert-relation-feature char name value
1456                                                 line-breaking
1457                                                 ccss readable))
1458               ;; ((memq name '(ideograph=
1459               ;;               original-ideograph-of
1460               ;;               ancient-ideograph-of
1461               ;;               vulgar-ideograph-of
1462               ;;               wrong-ideograph-of
1463               ;;               ;; simplified-ideograph-of
1464               ;;               ideographic-variants
1465               ;;               ;; ideographic-different-form-of
1466               ;;               ))
1467               ;;  (insert (format "(%-18s%s " name line-breaking))
1468               ;;  (setq lbs (concat "\n" (make-string (current-column) ?\ ))
1469               ;;        separator nil)
1470               ;;  (while (consp value)
1471               ;;    (setq cell (car value))
1472               ;;    (if (and (consp cell)
1473               ;;             (consp (car cell)))
1474               ;;        (progn
1475               ;;          (if separator
1476               ;;              (insert lbs))
1477               ;;          (char-db-insert-alist cell readable)
1478               ;;          (setq separator lbs))
1479               ;;      (if separator
1480               ;;          (insert separator))
1481               ;;      (insert (prin1-to-string cell))
1482               ;;      (setq separator " "))
1483               ;;    (setq value (cdr value)))
1484               ;;  (insert ")")
1485               ;;  (insert line-breaking))
1486               ((consp value)
1487                (insert (format "(%-18s " name))
1488                (setq lbs (concat "\n" (make-string (current-column) ?\ ))
1489                      separator nil)
1490                (while (consp value)
1491                  (setq cell (car value))
1492                  (if (and (consp cell)
1493                           (consp (car cell))
1494                           (setq ret (condition-case nil
1495                                         (find-char cell)
1496                                       (error nil))))
1497                      (progn
1498                        (setq rest cell
1499                              al nil
1500                              cal nil)
1501                        (while rest
1502                          (setq key (car (car rest)))
1503                          (if (find-charset key)
1504                              (setq cal (cons key cal))
1505                            (setq al (cons key al)))
1506                          (setq rest (cdr rest)))
1507                        (if separator
1508                            (insert lbs))
1509                        (insert-char-attributes ret
1510                                                readable
1511                                                al ; cal
1512                                                nil 'for-sub-node)
1513                        (setq separator lbs))
1514                    (setq ret (prin1-to-string cell))
1515                    (if separator
1516                        (if (< (+ (current-column)
1517                                  (length ret)
1518                                  (length separator))
1519                               76)
1520                            (insert separator)
1521                          (insert lbs)))
1522                    (insert ret)
1523                    (setq separator " "))
1524                  (setq value (cdr value)))
1525                (insert ")")
1526                (insert line-breaking))
1527               (t
1528                (insert (format "(%-18s" name))
1529                (setq ret (prin1-to-string value))
1530                (unless (< (+ (current-column)
1531                              (length ret)
1532                              3)
1533                           76)
1534                  (insert line-breaking))
1535                (insert " . " ret ")" line-breaking)
1536                ;; (insert (format "(%-18s . %S)%s"
1537                ;;                 name value
1538                ;;                 line-breaking))
1539                )
1540               ))
1541       (setq attributes (cdr attributes)))
1542     (insert ")")))
1543
1544 (defun insert-char-data (char &optional readable
1545                               attributes)
1546   (save-restriction
1547     (narrow-to-region (point)(point))
1548     (insert "(define-char
1549   '")
1550     (insert-char-attributes char readable attributes)
1551     (insert ")\n")
1552     (goto-char (point-min))
1553     (while (re-search-forward "[ \t]+$" nil t)
1554       (replace-match ""))
1555     ;; from tabify.
1556     (goto-char (point-min))
1557     (while (re-search-forward "[ \t][ \t][ \t]*" nil t)
1558       (let ((column (current-column))
1559             (indent-tabs-mode t))
1560         (delete-region (match-beginning 0) (point))
1561         (indent-to column)))
1562     (goto-char (point-max))
1563     ;; (tabify (point-min)(point-max))
1564     ))
1565
1566 (defun insert-char-data-with-variant (char &optional printable
1567                                            no-ucs-unified
1568                                            script excluded-script)
1569   (insert-char-data char printable)
1570   (let ((variants (char-variants char))
1571         rest
1572         variant vs ret)
1573     (setq variants (sort variants #'<))
1574     (setq rest variants)
1575     (setq variants (cons char variants))
1576     (while rest
1577       (setq variant (car rest))
1578       (unless (get-char-attribute variant '<-subsumptive)
1579         (if (and (or (null script)
1580                      (null (setq vs (get-char-attribute variant 'script)))
1581                      (memq script vs))
1582                  (or (null excluded-script)
1583                      (null (setq vs (get-char-attribute variant 'script)))
1584                      (not (memq excluded-script vs))))
1585             (unless (and no-ucs-unified (get-char-attribute variant '=ucs))
1586               (insert-char-data variant printable)
1587               (if (setq ret (char-variants variant))
1588                   (while ret
1589                     (or (memq (car ret) variants)
1590                         ;; (get-char-attribute (car ret) '<-subsumptive)
1591                         (setq rest (nconc rest (list (car ret)))))
1592                     (setq ret (cdr ret)))))))
1593       (setq rest (cdr rest)))))
1594
1595 (defun insert-char-range-data (min max &optional script excluded-script)
1596   (let ((code min)
1597         char)
1598     (while (<= code max)
1599       (setq char (decode-char '=ucs code))
1600       (if (encode-char char '=ucs 'defined-only)
1601           (insert-char-data-with-variant char nil 'no-ucs-unified
1602                                          script excluded-script))
1603       (setq code (1+ code)))))
1604
1605 (defun write-char-range-data-to-file (min max file
1606                                           &optional script excluded-script)
1607   (let ((coding-system-for-write char-db-file-coding-system))
1608     (with-temp-buffer
1609       (insert (format ";; -*- coding: %s -*-\n"
1610                       char-db-file-coding-system))
1611       (insert-char-range-data min max script excluded-script)
1612       (write-region (point-min)(point-max) file))))
1613
1614 (defvar what-character-original-window-configuration)
1615
1616 ;;;###autoload
1617 (defun what-char-definition (char)
1618   (interactive (list (char-after)))
1619   (let ((buf (get-buffer-create "*Character Description*"))
1620         (the-buf (current-buffer))
1621         (win-conf (current-window-configuration)))
1622     (pop-to-buffer buf)
1623     (make-local-variable 'what-character-original-window-configuration)
1624     (setq what-character-original-window-configuration win-conf)
1625     (setq buffer-read-only nil)
1626     (erase-buffer)
1627     (condition-case err
1628         (progn
1629           (insert-char-data-with-variant char 'printable)
1630           (unless (char-attribute-alist char)
1631             (insert (format ";; = %c\n"
1632                             (let* ((rest (split-char char))
1633                                    (ccs (pop rest))
1634                                    (code (pop rest)))
1635                               (while rest
1636                                 (setq code (logior (lsh code 8)
1637                                                    (pop rest))))
1638                               (decode-char ccs code)))))
1639           ;; (char-db-update-comment)
1640           (set-buffer-modified-p nil)
1641           (view-mode the-buf (lambda (buf)
1642                                (set-window-configuration
1643                                 what-character-original-window-configuration)
1644                                ))
1645           (goto-char (point-min)))
1646       (error (progn
1647                (set-window-configuration
1648                 what-character-original-window-configuration)
1649                (signal (car err) (cdr err)))))))
1650
1651
1652 ;;; @ end
1653 ;;;
1654
1655 (provide 'char-db-util)
1656
1657 ;;; char-db-util.el ends here