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-014
 392     =big5-cdp-var-001
 393     =big5-cdp-var-002
 394     =big5-cdp-var-003
 395     =big5-cdp-var-004
 396     =big5-cdp-var-005
 397     =big5-cdp-var-007
 398     =big5-cdp-var-010
 399     =big5-cdp-itaiji-001
 400     =big5-cdp-itaiji-002
 401     =big5-cdp-itaiji-003
 402     =big5-cdp-itaiji-004
 403     =big5-cdp-itaiji-005
 404     =>big5-cdp-itaiji-001
 405     =>big5-cdp-itaiji-002
 406     =>big5-cdp-itaiji-003
 407     =>big5-cdp-itaiji-004
 408     =>big5-cdp-var-001
 409     =>ucs@iwds-1
 410     =>ucs@cognate
 411     =>ucs@component
 412     =>ucs@iwds-1/normalized
 413     =>ucs-itaiji-001@iwds-1
 414     =>ucs-itaiji-002@iwds-1
 415     =>ucs-itaiji-003@iwds-1
 416     =>ucs-itaiji-006@iwds-1
 417     =>big5-cdp@component
 418     =>big5-cdp@cognate
 419     ==ucs@gb
 420     =ucs@gb
 421     ==ucs-var-002
 422     =ucs@JP/hanazono
 423     =koseki
 424     =shuowen-jiguge
 425     ===shuowen-jiguge5
 426     ===shuowen-jiguge4))
 427
 428
 429 ;;; @ char-db formatters
 430 ;;;
 431
 432 (defun char-db-make-char-spec (char)
 433   (let (ret char-spec)
 434     (cond ((characterp char)
 435            (cond ((and (setq ret (encode-char char '=ucs 'defined-only))
 436                        (not (and (<= #xE000 ret)(<= ret #xF8FF))))
 437                   (setq char-spec (list (cons '=ucs ret)))
 438                   (cond ((setq ret (get-char-attribute char 'name))
 439                          (setq char-spec (cons (cons 'name ret) char-spec))
 440                          )
 441                         ((setq ret (get-char-attribute char 'name*))
 442                          (setq char-spec (cons (cons 'name* ret) char-spec))
 443                          ))
 444                   )
 445                  ((encode-char char '=mj 'defined-only)
 446                   (setq char-spec nil)
 447                   (dolist (ccs (charset-list))
 448                     (if (and (or (eq ccs '=mj)
 449                                  ;; (eq (charset-property ccs 'iso-ir) 177)
 450                                  (string-match "=ucs@" (symbol-name ccs))
 451                                  )
 452                              (setq ccs (charset-name ccs))
 453                              (null (assq ccs char-spec))
 454                              (setq ret (encode-char char ccs 'defined-only)))
 455                         (setq char-spec (cons (cons ccs ret) char-spec))))
 456                   )
 457                  ((encode-char char '==mj 'defined-only)
 458                   (setq char-spec nil)
 459                   (dolist (ccs (charset-list))
 460                     (if (and (or (eq ccs '==mj)
 461                                  ;; (eq (charset-property ccs 'iso-ir) 177)
 462                                  (string-match "=ucs@" (symbol-name ccs))
 463                                  )
 464                              (setq ccs (charset-name ccs))
 465                              (null (assq ccs char-spec))
 466                              (setq ret (encode-char char ccs 'defined-only)))
 467                         (setq char-spec (cons (cons ccs ret) char-spec))))
 468                   )
 469                  ((encode-char char '=adobe-japan1 'defined-only)
 470                   (setq char-spec nil)
 471                   (dolist (ccs (charset-list))
 472                     (if (and (or (memq ccs
 473                                        '(=adobe-japan1-0
 474                                          =adobe-japan1-1
 475                                          =adobe-japan1-2
 476                                          =adobe-japan1-3
 477                                          =adobe-japan1-4
 478                                          =adobe-japan1-5
 479                                          =adobe-japan1-6
 480                                          ))
 481                                  ;; (eq (charset-property ccs 'iso-ir) 177)
 482                                  (string-match "=ucs@" (symbol-name ccs))
 483                                  )
 484                              (setq ccs (charset-name ccs))
 485                              (null (assq ccs char-spec))
 486                              (setq ret (encode-char char ccs 'defined-only)))
 487                         (setq char-spec (cons (cons ccs ret) char-spec))))
 488                   )
 489                  ((encode-char char '==adobe-japan1 'defined-only)
 490                   (setq char-spec nil)
 491                   (dolist (ccs (charset-list))
 492                     (if (and (or (memq ccs
 493                                        '(==adobe-japan1-0
 494                                          ==adobe-japan1-1
 495                                          ==adobe-japan1-2
 496                                          ==adobe-japan1-3
 497                                          ==adobe-japan1-4
 498                                          ==adobe-japan1-5
 499                                          ==adobe-japan1-6
 500                                          ))
 501                                  ;; (eq (charset-property ccs 'iso-ir) 177)
 502                                  (string-match "=ucs@" (symbol-name ccs))
 503                                  )
 504                              (setq ccs (charset-name ccs))
 505                              (null (assq ccs char-spec))
 506                              (setq ret (encode-char char ccs 'defined-only)))
 507                         (setq char-spec (cons (cons ccs ret) char-spec))))
 508                   )
 509                  ((setq ret
 510                         (catch 'tag
 511                           (let ((rest char-db-coded-charset-priority-list)
 512                                 ccs)
 513                             (while rest
 514                               (setq ccs (charset-name
 515                                          (find-charset (car rest))))
 516                               (if (setq ret
 517                                         (encode-char char ccs
 518                                                      'defined-only))
 519                                   (throw 'tag (cons ccs ret)))
 520                               (setq rest (cdr rest))))))
 521                   (setq char-spec (list ret))
 522                   (dolist (ccs (delq (car ret) (charset-list)))
 523                     (if (and (or (charset-iso-final-char ccs)
 524                                  (memq ccs
 525                                        '(=daikanwa
 526                                          =daikanwa@rev2
 527                                          ;; =gt-k
 528                                          =jis-x0208@1997
 529                                          ))
 530                                  (and (eq (charset-property ccs 'iso-ir) 177)
 531                                       (not (eq (charset-name ccs) '=ucs@big5)))
 532                                  ;; (string-match "=ucs@" (symbol-name ccs))
 533                                  )
 534                              (setq ccs (charset-name ccs))
 535                              (null (assq ccs char-spec))
 536                              (setq ret (encode-char char ccs 'defined-only)))
 537                         (setq char-spec (cons (cons ccs ret) char-spec))))
 538                   (if (null char-spec)
 539                       (setq char-spec (split-char char)))
 540                   (cond ((setq ret (get-char-attribute char 'name))
 541                          (setq char-spec (cons (cons 'name ret) char-spec))
 542                          )
 543                         ((setq ret (get-char-attribute char 'name*))
 544                          (setq char-spec (cons (cons 'name* ret) char-spec))
 545                          ))
 546                   )
 547                  ((setq ret (get-char-attribute
 548                              char 'ideographic-combination))
 549                   (setq char-spec
 550                         (cons (cons 'ideographic-combination ret)
 551                               char-spec))
 552                   ))
 553            char-spec)
 554           ((consp char)
 555            char))))
 556
 557 (defun char-db-insert-char-spec (char &optional readable column
 558                                       required-features)
 559   (unless column
 560     (setq column (current-column)))
 561   (let (char-spec temp-char)
 562     (setq char-spec (char-db-make-char-spec char))
 563     (unless (or (characterp char) ; char
 564                 (condition-case nil
 565                     (setq char (find-char char-spec))
 566                   (error nil)))
 567       ;; define temporary character
 568       ;;   Current implementation is dirty.
 569       (setq temp-char (define-char (cons '(ideograph-daikanwa . 0)
 570                                          char-spec)))
 571       (remove-char-attribute temp-char 'ideograph-daikanwa)
 572       (setq char temp-char))
 573     (insert-char-attributes char
 574                             readable
 575                             (union (mapcar #'car char-spec)
 576                                    required-features)
 577                             nil 'for-sub-node)
 578     (when temp-char
 579       ;; undefine temporary character
 580       ;;   Current implementation is dirty.
 581       (setq char-spec (char-attribute-alist temp-char))
 582       (while char-spec
 583         (remove-char-attribute temp-char (car (car char-spec)))
 584         (setq char-spec (cdr char-spec))))))
 585
 586 (defun char-db-insert-alist (alist &optional readable column)
 587   (unless column
 588     (setq column (current-column)))
 589   (let ((line-breaking
 590          (concat "\n" (make-string (1+ column) ?\ )))
 591         name value
 592         ret al ; cal
 593         key
 594         lbs cell rest separator)
 595     (insert "(")
 596     (while alist
 597       (setq name (car (car alist))
 598             value (cdr (car alist)))
 599       (cond ((eq name 'char)
 600              (insert "(char . ")
 601              (if (setq ret (condition-case nil
 602                                (find-char value)
 603                              (error nil)))
 604                  (progn
 605                    (setq al nil
 606                          ;; cal nil
 607                          )
 608                    (while value
 609                      (setq key (car (car value)))
 610                      ;; (if (find-charset key)
 611                      ;;     (setq cal (cons key cal))
 612                      (setq al (cons key al))
 613                      ;; )
 614                      (setq value (cdr value)))
 615                    (insert-char-attributes ret
 616                                            readable
 617                                            (or al 'none) ; cal
 618                                            nil 'for-sub-node))
 619                (insert (prin1-to-string value)))
 620              (insert ")")
 621              (insert line-breaking))
 622             ((consp value)
 623              (insert (format "(%-18s " name))
 624              (setq lbs (concat "\n" (make-string (current-column) ?\ )))
 625              (while (consp value)
 626                (setq cell (car value))
 627                (if (and (consp cell)
 628                         (consp (car cell))
 629                         (setq ret (condition-case nil
 630                                       (find-char cell)
 631                                     (error nil)))
 632                         )
 633                    (progn
 634                      (setq rest cell
 635                            al nil
 636                            ;; cal nil
 637                            )
 638                      (while rest
 639                        (setq key (car (car rest)))
 640                        ;; (if (find-charset key)
 641                        ;;     (setq cal (cons key cal))
 642                        (setq al (cons key al))
 643                        ;; )
 644                        (setq rest (cdr rest)))
 645                      (if separator
 646                          (insert lbs))
 647                      (insert-char-attributes ret
 648                                              readable
 649                                              al ; cal
 650                                              nil 'for-sub-node)
 651                      (setq separator lbs))
 652                  (if separator
 653                      (insert separator))
 654                  (insert (prin1-to-string cell))
 655                  (setq separator " "))
 656                (setq value (cdr value)))
 657              (insert ")")
 658              (insert line-breaking))
 659             (t
 660              (insert (format "(%-18s . %S)%s"
 661                              name value
 662                              line-breaking))))
 663       (setq alist (cdr alist))))
 664   (insert ")"))
 665
 666 (defun char-db-insert-char-reference (plist &optional readable column)
 667   (unless column
 668     (setq column (current-column)))
 669   (let ((line-breaking
 670          (concat "\n" (make-string (1+ column) ?\ )))
 671         (separator "")
 672         name value)
 673     (insert "(")
 674     (while plist
 675       (setq name (pop plist))
 676       (setq value (pop plist))
 677       (cond ((eq name :char)
 678              (insert separator)
 679              (insert ":char\t")
 680              (cond ((numberp value)
 681                     (setq value (decode-char '=ucs value)))
 682                    ;; ((consp value)
 683                    ;;  (setq value (or (find-char value)
 684                    ;;                  value)))
 685                    )
 686              (char-db-insert-char-spec value readable)
 687              (insert line-breaking)
 688              (setq separator ""))
 689             ((eq name :radical)
 690              (insert (format "%s%s\t%d ; %c%s"
 691                              separator
 692                              name value
 693                              (ideographic-radical value)
 694                              line-breaking))
 695              (setq separator ""))
 696             (t
 697              (insert (format "%s%s\t%S" separator name value))
 698              (setq separator line-breaking)))
 699       ))
 700   (insert ")"))
 701
 702 (defun char-db-decode-isolated-char (ccs code-point)
 703   (let (ret)
 704     (setq ret
 705           (cond ((eq ccs 'arabic-iso8859-6)
 706                  (decode-char ccs code-point))
 707                 ;; ((eq ccs '=gt)
 708                 ;;  (decode-builtin-char '==gt code-point))
 709                 ((and (memq ccs '(=gt-pj-1
 710                                   =gt-pj-2
 711                                   =gt-pj-3
 712                                   =gt-pj-4
 713                                   =gt-pj-5
 714                                   =gt-pj-6
 715                                   =gt-pj-7
 716                                   =gt-pj-8
 717                                   =gt-pj-9
 718                                   =gt-pj-10
 719                                   =gt-pj-11))
 720                       (setq ret (decode-char ccs code-point))
 721                       (setq ret (encode-char ret '=gt 'defined-only)))
 722                  (decode-builtin-char '=gt ret))
 723                 (t
 724                  (decode-builtin-char ccs code-point))))
 725     (cond ((null ret)
 726            (or (decode-char ccs code-point)
 727                (define-char (list (cons ccs code-point)))))
 728           ((and (<= 0 (char-int ret))
 729                 (<= (char-int ret) #x1F))
 730            (decode-char '=ucs (+ #x2400 (char-int ret))))
 731           ((= (char-int ret) #x7F)
 732            ?\u2421)
 733           (t ret))))
 734
 735 (defvar char-db-convert-obsolete-format t)
 736
 737 (defun char-db-insert-ccs-feature (name value line-breaking)
 738   (cond
 739    ((integerp value)
 740     (insert
 741      (format
 742       (cond
 743        ((memq name '(=>iwds-1
 744                      ==shinjigen
 745                      =shinjigen
 746                      =shinjigen@1ed ==shinjigen@1ed
 747                      =shinjigen@rev ==shinjigen@rev
 748                      =shinjigen/+p@rev ==shinjigen/+p@rev
 749                      ===daikanwa/ho ==daikanwa/ho
 750                      =daikanwa/ho =>>daikanwa/ho =>daikanwa/ho))
 751         "(%-18s .  %04d)\t; %c")
 752        ((eq name '=shinjigen@1ed/24pr)
 753         "(%-18s . %04d)\t; %c")
 754        ((or
 755          (memq name
 756                '(===daikanwa
 757                  ==daikanwa =daikanwa =>>daikanwa =>daikanwa
 758                  =daikanwa@rev1 =daikanwa@rev2
 759                  =daikanwa/+p ==daikanwa/+p ===daikanwa/+p
 760                  =>>daikanwa/+p
 761                  =daikanwa/+2p =>>daikanwa/+2p
 762                  =gt ==gt ===gt
 763                  =>>gt =+>gt =>gt
 764                  =gt-k ==gt-k ===gt-k
 765                  =>>gt-k =>gt-k
 766                  =adobe-japan1-0 ==adobe-japan1-0 ===adobe-japan1-0
 767                  =adobe-japan1-1 ==adobe-japan1-1 ===adobe-japan1-1
 768                  =adobe-japan1-2 ==adobe-japan1-2 ===adobe-japan1-2
 769                  =adobe-japan1-3 ==adobe-japan1-3 ===adobe-japan1-3
 770                  =adobe-japan1-4 ==adobe-japan1-4 ===adobe-japan1-4
 771                  =adobe-japan1-5 ==adobe-japan1-5 ===adobe-japan1-5
 772                  =adobe-japan1-6 ==adobe-japan1-6 ===adobe-japan1-6
 773                  =>>adobe-japan1-0 =+>adobe-japan1-0
 774                  =>>adobe-japan1-1 =+>adobe-japan1-1
 775                  =>>adobe-japan1-2 =+>adobe-japan1-2
 776                  =>>adobe-japan1-3 =+>adobe-japan1-3
 777                  =>>adobe-japan1-4 =+>adobe-japan1-4
 778                  =>>adobe-japan1-5 =+>adobe-japan1-5
 779                  =>>adobe-japan1-6 =+>adobe-japan1-6
 780                  =>cbeta =cbeta =>>cbeta ==cbeta ===cbeta
 781                  =zinbun-oracle =>zinbun-oracle
 782                  =daijiten ==daijiten ===daijiten ==>daijiten
 783                  ===hng-jou ===hng-keg ===hng-dng ===hng-mam
 784                  ===hng-drt ===hng-kgk ===hng-myz ===hng-kda
 785                  ===hng-khi ===hng-khm ===hng-fhs ===hng-hok
 786                  ===hng-kyd ===hng-sok
 787                  ===hng-yhk ===hng-kak ===hng-kar ===hng-kae
 788                  ===hng-sys ===hng-tsu ===hng-tzj
 789                  ===hng-hos ===hng-kkh ===hng-nak ===hng-jhk
 790                  ===hng-hod ===hng-gok ===hng-ink ===hng-nto
 791                  ===hng-nkm ===hng-k24 ===hng-ini ===hng-nkk
 792                  ===hng-kcc ===hng-kcj ===hng-kbk ===hng-sik
 793                  ===hng-skk ===hng-kyu ===hng-ksk ===hng-wan
 794                  ===hng-okd ===hng-wad ===hng-kmi
 795                  ===hng-zkd ===hng-doh ===hng-jyu ===hng-tzs
 796                  ===hng-sai ===hng-kad ===hng-kss
 797                  ===hng-kyo ===hng-ykk ===hng-saa
 798                  ===hng-sab ===hng-wks ===hng-wke ===hng-smk
 799                  ===hng-sgs ===hng-sts ===hng-kkd
 800                  =shuowen-jiguge ===shuowen-jiguge4 ===shuowen-jiguge5))
 801          ;; (string-match "^=adobe-" (symbol-name name))
 802          )
 803         "(%-18s . %05d)\t; %c")
 804        ((memq name '(=hanyo-denshi/ks
 805                      ==hanyo-denshi/ks ===hanyo-denshi/ks
 806                      =>>hanyo-denshi/ks
 807                      =koseki ==koseki
 808                      =mj ==mj ===mj =>>mj =>mj
 809                      =zihai mojikyo))
 810         "(%-18s . %06d)\t; %c")
 811        ((memq name '(=hanyo-denshi/tk =>hanyo-denshi/tk ==hanyo-denshi/tk))
 812         "(%-18s . %08d)\t; %c")
 813        ((>= (charset-dimension name) 2)
 814         "(%-18s . #x%04X)\t; %c")
 815        (t
 816         "(%-18s . #x%02X)\t; %c"))
 817       name
 818       (if (= (charset-iso-graphic-plane name) 1)
 819           (logior value
 820                   (cond ((= (charset-dimension name) 1)
 821                          #x80)
 822                         ((= (charset-dimension name) 2)
 823                          #x8080)
 824                         ((= (charset-dimension name) 3)
 825                          #x808080)
 826                         (t 0)))
 827         value)
 828       (char-db-decode-isolated-char name value)))
 829     (if (and (= (charset-chars name) 94)
 830              (= (charset-dimension name) 2))
 831         (insert (format " [%02d-%02d]"
 832                         (- (lsh value -8) 32)
 833                         (- (logand value 255) 32))))
 834     )
 835    (t
 836     (insert (format "(%-18s . %s)" name value))
 837     ))
 838   (insert line-breaking))
 839
 840 (defun char-db-insert-relation-feature (char name value line-breaking
 841                                              ccss readable)
 842   (insert (format "(%-18s%s " name line-breaking))
 843   (let ((lbs (concat "\n" (make-string (current-column) ?\ )))
 844         separator cell sources required-features
 845         ret)
 846     (while (consp value)
 847       (setq cell (car value))
 848       (if (integerp cell)
 849           (setq cell (decode-char '=ucs cell)))
 850       (cond
 851        ((eq name '->subsumptive)
 852         (when (or (not (some (lambda (atr)
 853                                (get-char-attribute cell atr))
 854                              char-db-ignored-attributes))
 855                   (some (lambda (ccs)
 856                           (encode-char cell ccs 'defined-only))
 857                         ccss))
 858           (if separator
 859               (insert lbs))
 860           (let ((char-db-ignored-attributes
 861                  (cons '<-subsumptive
 862                        char-db-ignored-attributes)))
 863             (insert-char-attributes cell readable nil nil 'for-sub-node))
 864           (setq separator lbs))
 865         )
 866        ((characterp cell)
 867         (setq sources
 868               (get-char-attribute
 869                char (intern (format "%s*sources" name))))
 870         (setq required-features nil)
 871         (dolist (source sources)
 872           (cond
 873            ((memq source '(JP
 874                            JP/Jouyou
 875                            shinjigen shinjigen@1ed shinjigen@rev))
 876             (setq required-features
 877                   (union required-features
 878                          '(=jis-x0208
 879                            =jis-x0208@1990
 880                            =jis-x0213-1@2000
 881                            =jis-x0213-1@2004
 882                            =jis-x0213-2
 883                            =jis-x0212
 884                            =jis-x0208@1983
 885                            =jis-x0208@1978
 886                            =shinjigen
 887                            =shinjigen@1ed
 888                            =shinjigen@rev
 889                            =shinjigen/+p@rev))))
 890            ((eq source 'CN)
 891             (setq required-features
 892                   (union required-features
 893                          '(=gb2312
 894                            =gb12345
 895                            =iso-ir165)))))
 896           (cond
 897            ((find-charset (setq ret (intern (format "=%s" source))))
 898             (setq required-features
 899                   (cons ret required-features)))
 900            (t (setq required-features
 901                     (cons source required-features)))))
 902         (cond ((string-match "@JP" (symbol-name name))
 903                (setq required-features
 904                      (union required-features
 905                             '(=jis-x0208
 906                               =jis-x0208@1990
 907                               =jis-x0213-1-2000
 908                               =jis-x0213-2-2000
 909                               =jis-x0212
 910                               =jis-x0208@1983
 911                               =jis-x0208@1978))))
 912               ((string-match "@CN" (symbol-name name))
 913                (setq required-features
 914                      (union required-features
 915                             '(=gb2312
 916                               =gb12345
 917                               =iso-ir165)))))
 918         (if separator
 919             (insert lbs))
 920         (if readable
 921             (insert (format "%S" cell))
 922           (char-db-insert-char-spec cell readable
 923                                     nil
 924                                     required-features))
 925         (setq separator lbs))
 926        ((consp cell)
 927         (if separator
 928             (insert lbs))
 929         (if (consp (car cell))
 930             (char-db-insert-char-spec cell readable)
 931           (char-db-insert-char-reference cell readable))
 932         (setq separator lbs))
 933        (t
 934         (if separator
 935             (insert separator))
 936         (insert (prin1-to-string cell))
 937         (setq separator " ")))
 938       (setq value (cdr value)))
 939     (insert ")")
 940     (insert line-breaking)))
 941
 942 (defun insert-char-attributes (char &optional readable attributes column
 943                                     for-sub-node)
 944   (unless column
 945     (setq column (current-column)))
 946   (let (name value ; has-long-ccs-name
 947         rest
 948         radical strokes
 949         (line-breaking
 950          (concat "\n" (make-string (1+ column) ?\ )))
 951         lbs cell separator ret
 952         key al cal
 953         dest-ccss ; sources required-features
 954         ccss)
 955     (let (atr-d)
 956       (setq attributes
 957             (sort (if attributes
 958                       (if (consp attributes)
 959                           (progn
 960                             (dolist (name attributes)
 961                               (unless (memq name char-db-ignored-attributes)
 962                                 (if (find-charset name)
 963                                     (push name ccss))
 964                                 (push name atr-d)))
 965                             atr-d))
 966                     (dolist (name (char-attribute-list))
 967                       (unless (memq name char-db-ignored-attributes)
 968                         (if (find-charset name)
 969                             (push name ccss))
 970                         (push name atr-d)))
 971                     atr-d)
 972                   #'char-attribute-name<)))
 973     (insert "(")
 974     (when (memq '<-subsumptive attributes)
 975       (when (or readable (not for-sub-node))
 976         (when (setq value (get-char-attribute char '<-subsumptive))
 977           (char-db-insert-relation-feature char '<-subsumptive value
 978                                            line-breaking
 979                                            ccss readable)))
 980       (setq attributes (delq '<-subsumptive attributes)))
 981     (when (and (memq '<-denotational attributes)
 982                (setq value (get-char-attribute char '<-denotational)))
 983       (char-db-insert-relation-feature char '<-denotational value
 984                                        line-breaking
 985                                        ccss readable)
 986       (setq attributes (delq '<-denotational attributes)))
 987     (when (and (memq '<-denotational@component attributes)
 988                (setq value (get-char-attribute char '<-denotational@component)))
 989       (char-db-insert-relation-feature char '<-denotational@component value
 990                                        line-breaking
 991                                        ccss readable)
 992       (setq attributes (delq '<-denotational@component attributes)))
 993     (when (and (memq '<-denotational@usage attributes)
 994                (setq value (get-char-attribute char '<-denotational@usage)))
 995       (char-db-insert-relation-feature char '<-denotational@usage value
 996                                        line-breaking
 997                                        ccss readable)
 998       (setq attributes (delq '<-denotational@usage attributes)))
 999     (when (and (memq 'name attributes)
1000                (setq value (get-char-attribute char 'name)))
1001       (insert (format
1002                (if (> (+ (current-column) (length value)) 48)
1003                    "(name . %S)%s"
1004                  "(name               . %S)%s")
1005                value line-breaking))
1006       (setq attributes (delq 'name attributes))
1007       )
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 'script attributes)
1018                (setq value (get-char-attribute char 'script)))
1019       (insert (format "(script\t\t%s)%s"
1020                       (mapconcat (function prin1-to-string)
1021                                  value " ")
1022                       line-breaking))
1023       (setq attributes (delq 'script attributes))
1024       )
1025     (dolist (name '(=>ucs =>ucs*))
1026       (when (and (memq name attributes)
1027                  (setq value (get-char-attribute char name)))
1028         (insert (format "(%-18s . #x%04X)\t; %c%s"
1029                         name value (decode-char '=ucs value)
1030                         line-breaking))
1031         (setq attributes (delq name attributes))))
1032     (dolist (name '(=>ucs@gb =>ucs@big5))
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
1037                         (decode-char (intern
1038                                       (concat "="
1039                                               (substring
1040                                                (symbol-name name) 2)))
1041                                      value)
1042                         line-breaking))
1043         (setq attributes (delq name attributes))
1044         ))
1045     ;; (dolist (name '(=>daikanwa))
1046     ;;   (when (and (memq name attributes)
1047     ;;              (setq value (get-char-attribute char name)))
1048     ;;     (insert
1049     ;;      (if (integerp value)
1050     ;;          (format "(%-18s . %05d)\t; %c%s"
1051     ;;                  name value (decode-char '=daikanwa value)
1052     ;;                  line-breaking)
1053     ;;        (format "(%-18s %s)\t; %c%s"
1054     ;;                name
1055     ;;                (mapconcat (function prin1-to-string)
1056     ;;                           value " ")
1057     ;;                (char-representative-of-daikanwa char)
1058     ;;                line-breaking)))
1059     ;;     (setq attributes (delq name attributes))))
1060     (when (and (memq 'general-category attributes)
1061                (setq value (get-char-attribute char 'general-category)))
1062       (insert (format
1063                "(general-category\t%s) ; %s%s"
1064                (mapconcat (lambda (cell)
1065                             (format "%S" cell))
1066                           value " ")
1067                (cond ((rassoc value unidata-normative-category-alist)
1068                       "Normative Category")
1069                      ((rassoc value unidata-informative-category-alist)
1070                       "Informative Category")
1071                      (t
1072                       "Unknown Category"))
1073                line-breaking))
1074       (setq attributes (delq 'general-category attributes))
1075       )
1076     (when (and (memq 'bidi-category attributes)
1077                (setq value (get-char-attribute char 'bidi-category)))
1078       (insert (format "(bidi-category\t. %S)%s"
1079                       value
1080                       line-breaking))
1081       (setq attributes (delq 'bidi-category attributes))
1082       )
1083     (unless (or (not (memq 'mirrored attributes))
1084                 (eq (setq value (get-char-attribute char 'mirrored 'empty))
1085                     'empty))
1086       (insert (format "(mirrored\t\t. %S)%s"
1087                       value
1088                       line-breaking))
1089       (setq attributes (delq 'mirrored attributes))
1090       )
1091     (cond
1092      ((and (memq 'decimal-digit-value attributes)
1093            (setq value (get-char-attribute char 'decimal-digit-value)))
1094       (insert (format "(decimal-digit-value . %S)%s"
1095                       value
1096                       line-breaking))
1097       (setq attributes (delq 'decimal-digit-value attributes))
1098       (when (and (memq 'digit-value attributes)
1099                  (setq value (get-char-attribute char 'digit-value)))
1100         (insert (format "(digit-value\t . %S)%s"
1101                         value
1102                         line-breaking))
1103         (setq attributes (delq 'digit-value attributes))
1104         )
1105       (when (and (memq 'numeric-value attributes)
1106                  (setq value (get-char-attribute char 'numeric-value)))
1107         (insert (format "(numeric-value\t . %S)%s"
1108                         value
1109                         line-breaking))
1110         (setq attributes (delq 'numeric-value attributes))
1111         )
1112       )
1113      (t
1114       (when (and (memq 'digit-value attributes)
1115                  (setq value (get-char-attribute char 'digit-value)))
1116         (insert (format "(digit-value\t. %S)%s"
1117                         value
1118                         line-breaking))
1119         (setq attributes (delq 'digit-value attributes))
1120         )
1121       (when (and (memq 'numeric-value attributes)
1122                  (setq value (get-char-attribute char 'numeric-value)))
1123         (insert (format "(numeric-value\t. %S)%s"
1124                         value
1125                         line-breaking))
1126         (setq attributes (delq 'numeric-value attributes))
1127         )))
1128     (when (and (memq 'iso-10646-comment attributes)
1129                (setq value (get-char-attribute char 'iso-10646-comment)))
1130       (insert (format "(iso-10646-comment\t. %S)%s"
1131                       value
1132                       line-breaking))
1133       (setq attributes (delq 'iso-10646-comment attributes))
1134       )
1135     (when (and (memq 'morohashi-daikanwa attributes)
1136                (setq value (get-char-attribute char 'morohashi-daikanwa)))
1137       (insert (format "(morohashi-daikanwa\t%s)%s"
1138                       (mapconcat (function prin1-to-string) value " ")
1139                       line-breaking))
1140       (setq attributes (delq 'morohashi-daikanwa attributes))
1141       )
1142     (setq radical nil
1143           strokes nil)
1144     (when (and (memq 'ideographic-radical attributes)
1145                (setq value (get-char-attribute char 'ideographic-radical)))
1146       (setq radical value)
1147       (insert (format "(ideographic-radical . %S)\t; %c%s"
1148                       radical
1149                       (ideographic-radical radical)
1150                       line-breaking))
1151       (setq attributes (delq 'ideographic-radical attributes))
1152       )
1153     (when (and (memq 'shuowen-radical attributes)
1154                (setq value (get-char-attribute char 'shuowen-radical)))
1155       (insert (format "(shuowen-radical\t. %S)\t; %c%s"
1156                       value
1157                       (shuowen-radical value)
1158                       line-breaking))
1159       (setq attributes (delq 'shuowen-radical attributes))
1160       )
1161     (let (key)
1162       (dolist (domain
1163                (append
1164                 char-db-feature-domains
1165                 (let (dest domain)
1166                   (dolist (feature (char-attribute-list))
1167                     (setq feature (symbol-name feature))
1168                     (when (string-match
1169                            "\\(radical\\|strokes\\)@\\([^@*]+\\)\\(\\*\\|$\\)"
1170                            feature)
1171                       (setq domain (intern (match-string 2 feature)))
1172                      (unless (memq domain dest)
1173                        (setq dest (cons domain dest)))))
1174                   (sort dest #'string<))))
1175         (setq key (intern (format "%s@%s" 'ideographic-radical domain)))
1176         (when (and (memq key attributes)
1177                    (setq value (get-char-attribute char key)))
1178           (setq radical value)
1179           (insert (format "(%s . %S)\t; %c%s"
1180                           key
1181                           radical
1182                           (ideographic-radical radical)
1183                           line-breaking))
1184           (setq attributes (delq key attributes))
1185           )
1186         (setq key (intern (format "%s@%s" 'ideographic-strokes domain)))
1187         (when (and (memq key attributes)
1188                    (setq value (get-char-attribute char key)))
1189           (setq strokes value)
1190           (insert (format "(%s . %S)%s"
1191                           key
1192                           strokes
1193                           line-breaking))
1194           (setq attributes (delq key attributes))
1195           )
1196         (setq key (intern (format "%s@%s" 'total-strokes domain)))
1197         (when (and (memq key attributes)
1198                    (setq value (get-char-attribute char key)))
1199           (insert (format "(%s       . %S)%s"
1200                           key
1201                           value
1202                           line-breaking))
1203           (setq attributes (delq key attributes))
1204           )
1205         (dolist (feature '(ideographic-radical
1206                            ideographic-strokes
1207                            total-strokes))
1208           (setq key (intern (format "%s@%s*sources" feature domain)))
1209           (when (and (memq key attributes)
1210                      (setq value (get-char-attribute char key)))
1211             (insert (format "(%s%s" key line-breaking))
1212             (dolist (cell value)
1213               (insert (format " %s" cell)))
1214             (insert ")")
1215             (insert line-breaking)
1216             (setq attributes (delq key attributes))
1217             ))
1218         ))
1219     (when (and (memq 'ideographic-strokes attributes)
1220                (setq value (get-char-attribute char 'ideographic-strokes)))
1221       (setq strokes value)
1222       (insert (format "(ideographic-strokes . %S)%s"
1223                       strokes
1224                       line-breaking))
1225       (setq attributes (delq 'ideographic-strokes attributes))
1226       )
1227     (when (and (memq 'kangxi-radical attributes)
1228                (setq value (get-char-attribute char 'kangxi-radical)))
1229       (unless (eq value radical)
1230         (insert (format "(kangxi-radical\t . %S)\t; %c%s"
1231                         value
1232                         (ideographic-radical value)
1233                         line-breaking))
1234         (or radical
1235             (setq radical value)))
1236       (setq attributes (delq 'kangxi-radical attributes))
1237       )
1238     (when (and (memq 'kangxi-strokes attributes)
1239                (setq value (get-char-attribute char 'kangxi-strokes)))
1240       (unless (eq value strokes)
1241         (insert (format "(kangxi-strokes\t . %S)%s"
1242                         value
1243                         line-breaking))
1244         (or strokes
1245             (setq strokes value)))
1246       (setq attributes (delq 'kangxi-strokes attributes))
1247       )
1248     ;; (when (and (memq 'japanese-radical attributes)
1249     ;;            (setq value (get-char-attribute char 'japanese-radical)))
1250     ;;   (unless (eq value radical)
1251     ;;     (insert (format "(japanese-radical\t . %S)\t; %c%s"
1252     ;;                     value
1253     ;;                     (ideographic-radical value)
1254     ;;                     line-breaking))
1255     ;;     (or radical
1256     ;;         (setq radical value)))
1257     ;;   (setq attributes (delq 'japanese-radical attributes))
1258     ;;   )
1259     (when (and (memq 'japanese-strokes attributes)
1260                (setq value (get-char-attribute char 'japanese-strokes)))
1261       (unless (eq value strokes)
1262         (insert (format "(japanese-strokes\t . %S)%s"
1263                         value
1264                         line-breaking))
1265         (or strokes
1266             (setq strokes value)))
1267       (setq attributes (delq 'japanese-strokes attributes))
1268       )
1269     (when (and (memq 'cns-radical attributes)
1270                (setq value (get-char-attribute char 'cns-radical)))
1271       (insert (format "(cns-radical\t . %S)\t; %c%s"
1272                       value
1273                       (ideographic-radical value)
1274                       line-breaking))
1275       (setq attributes (delq 'cns-radical attributes))
1276       )
1277     (when (and (memq 'cns-strokes attributes)
1278                (setq value (get-char-attribute char 'cns-strokes)))
1279       (unless (eq value strokes)
1280         (insert (format "(cns-strokes\t . %S)%s"
1281                         value
1282                         line-breaking))
1283         (or strokes
1284             (setq strokes value)))
1285       (setq attributes (delq 'cns-strokes attributes))
1286       )
1287     ;; (when (and (memq 'ideographic- attributes)
1288     ;;            (setq value (get-char-attribute char 'ideographic-)))
1289     ;;   (insert "(ideographic-       ")
1290     ;;   (setq lbs (concat "\n" (make-string (current-column) ?\ ))
1291     ;;         separator nil)
1292     ;;   (while (consp value)
1293     ;;     (setq cell (car value))
1294     ;;     (if (integerp cell)
1295     ;;         (setq cell (decode-char '=ucs cell)))
1296     ;;     (cond ((characterp cell)
1297     ;;            (if separator
1298     ;;                (insert lbs))
1299     ;;            (if readable
1300     ;;                (insert (format "%S" cell))
1301     ;;              (char-db-insert-char-spec cell readable))
1302     ;;            (setq separator lbs))
1303     ;;           ((consp cell)
1304     ;;            (if separator
1305     ;;                (insert lbs))
1306     ;;            (if (consp (car cell))
1307     ;;                (char-db-insert-char-spec cell readable)
1308     ;;              (char-db-insert-char-reference cell readable))
1309     ;;            (setq separator lbs))
1310     ;;           (t
1311     ;;            (if separator
1312     ;;                (insert separator))
1313     ;;            (insert (prin1-to-string cell))
1314     ;;            (setq separator " ")))
1315     ;;     (setq value (cdr value)))
1316     ;;   (insert ")")
1317     ;;   (insert line-breaking)
1318     ;;   (setq attributes (delq 'ideographic- attributes)))
1319     (when (and (memq 'total-strokes attributes)
1320                (setq value (get-char-attribute char 'total-strokes)))
1321       (insert (format "(total-strokes       . %S)%s"
1322                       value
1323                       line-breaking))
1324       (setq attributes (delq 'total-strokes attributes))
1325       )
1326     ;; (when (and (memq '->ideograph attributes)
1327     ;;            (setq value (get-char-attribute char '->ideograph)))
1328     ;;   (insert (format "(->ideograph\t%s)%s"
1329     ;;                   (mapconcat (lambda (code)
1330     ;;                                (cond ((symbolp code)
1331     ;;                                       (symbol-name code))
1332     ;;                                      ((integerp code)
1333     ;;                                       (format "#x%04X" code))
1334     ;;                                      (t
1335     ;;                                       (format "%s %S"
1336     ;;                                               line-breaking code))))
1337     ;;                              value " ")
1338     ;;                   line-breaking))
1339     ;;   (setq attributes (delq '->ideograph attributes))
1340     ;;   )
1341     ;; (when (and (memq '->decomposition attributes)
1342     ;;            (setq value (get-char-attribute char '->decomposition)))
1343     ;;   (insert (format "(->decomposition\t%s)%s"
1344     ;;                   (mapconcat (lambda (code)
1345     ;;                                (cond ((symbolp code)
1346     ;;                                       (symbol-name code))
1347     ;;                                      ((characterp code)
1348     ;;                                       (if readable
1349     ;;                                           (format "%S" code)
1350     ;;                                         (format "#x%04X"
1351     ;;                                                 (char-int code))
1352     ;;                                         ))
1353     ;;                                      ((integerp code)
1354     ;;                                       (format "#x%04X" code))
1355     ;;                                      (t
1356     ;;                                       (format "%s%S" line-breaking code))))
1357     ;;                              value " ")
1358     ;;                   line-breaking))
1359     ;;   (setq attributes (delq '->decomposition attributes))
1360     ;;   )
1361     (if (equal (get-char-attribute char '->titlecase)
1362                (get-char-attribute char '->uppercase))
1363         (setq attributes (delq '->titlecase attributes)))
1364     (when (and (memq '->mojikyo attributes)
1365                (setq value (get-char-attribute char '->mojikyo)))
1366       (insert (format "(->mojikyo\t\t. %06d)\t; %c%s"
1367                       value (decode-char 'mojikyo value)
1368                       line-breaking))
1369       (setq attributes (delq '->mojikyo attributes))
1370       )
1371     (unless readable
1372       (dolist (ignored '(composition
1373                          ->denotational <-subsumptive ->ucs-unified
1374                          ;; ->ideographic-component-forms
1375                          ))
1376         (setq attributes (delq ignored attributes))))
1377     (while attributes
1378       (setq name (car attributes))
1379       (unless (eq (setq value (get-char-attribute char name 'value-is-empty))
1380                   'value-is-empty)
1381         (cond ((setq ret (find-charset name))
1382                (setq name (charset-name ret))
1383                (when (not (memq name dest-ccss))
1384                  (setq dest-ccss (cons name dest-ccss))
1385                  (char-db-insert-ccs-feature name value line-breaking))
1386                )
1387               ((string-match "^=>ucs@" (symbol-name name))
1388                (insert (format "(%-18s . #x%04X)\t; %c%s"
1389                                name value (decode-char '=ucs value)
1390                                line-breaking))
1391                )
1392               ((eq name 'jisx0208-1978/4X)
1393                (insert (format "(%-18s . #x%04X)%s"
1394                                name value
1395                                line-breaking))
1396                )
1397               ((and
1398                 (not readable)
1399                 (not (eq name '->subsumptive))
1400                 (not (eq name '->uppercase))
1401                 (not (eq name '->lowercase))
1402                 (not (eq name '->titlecase))
1403                 (not (eq name '->canonical))
1404                 (not (eq name '->Bopomofo))
1405                 (not (eq name '->mistakable))
1406                 (not (eq name '->ideographic-variants))
1407                 (null (get-char-attribute
1408                        char (intern (format "%s*sources" name))))
1409                 (not (string-match "\\*sources$" (symbol-name name)))
1410                 (null (get-char-attribute
1411                        char (intern (format "%s*note" name))))
1412                 (not (string-match "\\*note$" (symbol-name name)))
1413                 (or (eq name '<-identical)
1414                     (eq name '<-uppercase)
1415                     (eq name '<-lowercase)
1416                     (eq name '<-titlecase)
1417                     (eq name '<-canonical)
1418                     (eq name '<-ideographic-variants)
1419                     ;; (eq name '<-synonyms)
1420                     (string-match "^<-synonyms" (symbol-name name))
1421                     (eq name '<-mistakable)
1422                     (when (string-match "^->" (symbol-name name))
1423                       (cond
1424                        ((string-match "^->fullwidth" (symbol-name name))
1425                         (not (and (consp value)
1426                                   (characterp (car value))
1427                                   (encode-char
1428                                    (car value) '=ucs 'defined-only)))
1429                         )
1430                        (t)))
1431                     ))
1432                )
1433               ((or (eq name 'ideographic-structure)
1434                    (eq name 'ideographic-structure@apparent)
1435                    (eq name 'ideographic-structure@apparent/leftmost)
1436                    (eq name 'ideographic-combination)
1437                    ;; (eq name 'ideographic-)
1438                    (eq name '=decomposition)
1439                    (char-feature-base-name= '=decomposition name)
1440                    (char-feature-base-name= '=>decomposition name)
1441                    ;; (string-match "^=>*decomposition\\(@[^*]+\\)?$"
1442                    ;;               (symbol-name name))
1443                    (string-match "^\\(->\\|<-\\)[^*]*$" (symbol-name name))
1444                    (string-match "^\\(->\\|<-\\)[^*]*\\*sources$"
1445                                  (symbol-name name))
1446                    )
1447                (char-db-insert-relation-feature char name value
1448                                                 line-breaking
1449                                                 ccss readable))
1450               ;; ((memq name '(ideograph=
1451               ;;               original-ideograph-of
1452               ;;               ancient-ideograph-of
1453               ;;               vulgar-ideograph-of
1454               ;;               wrong-ideograph-of
1455               ;;               ;; simplified-ideograph-of
1456               ;;               ideographic-variants
1457               ;;               ;; ideographic-different-form-of
1458               ;;               ))
1459               ;;  (insert (format "(%-18s%s " name line-breaking))
1460               ;;  (setq lbs (concat "\n" (make-string (current-column) ?\ ))
1461               ;;        separator nil)
1462               ;;  (while (consp value)
1463               ;;    (setq cell (car value))
1464               ;;    (if (and (consp cell)
1465               ;;             (consp (car cell)))
1466               ;;        (progn
1467               ;;          (if separator
1468               ;;              (insert lbs))
1469               ;;          (char-db-insert-alist cell readable)
1470               ;;          (setq separator lbs))
1471               ;;      (if separator
1472               ;;          (insert separator))
1473               ;;      (insert (prin1-to-string cell))
1474               ;;      (setq separator " "))
1475               ;;    (setq value (cdr value)))
1476               ;;  (insert ")")
1477               ;;  (insert line-breaking))
1478               ((consp value)
1479                (insert (format "(%-18s " name))
1480                (setq lbs (concat "\n" (make-string (current-column) ?\ ))
1481                      separator nil)
1482                (while (consp value)
1483                  (setq cell (car value))
1484                  (if (and (consp cell)
1485                           (consp (car cell))
1486                           (setq ret (condition-case nil
1487                                         (find-char cell)
1488                                       (error nil))))
1489                      (progn
1490                        (setq rest cell
1491                              al nil
1492                              cal nil)
1493                        (while rest
1494                          (setq key (car (car rest)))
1495                          (if (find-charset key)
1496                              (setq cal (cons key cal))
1497                            (setq al (cons key al)))
1498                          (setq rest (cdr rest)))
1499                        (if separator
1500                            (insert lbs))
1501                        (insert-char-attributes ret
1502                                                readable
1503                                                al ; cal
1504                                                nil 'for-sub-node)
1505                        (setq separator lbs))
1506                    (setq ret (prin1-to-string cell))
1507                    (if separator
1508                        (if (< (+ (current-column)
1509                                  (length ret)
1510                                  (length separator))
1511                               76)
1512                            (insert separator)
1513                          (insert lbs)))
1514                    (insert ret)
1515                    (setq separator " "))
1516                  (setq value (cdr value)))
1517                (insert ")")
1518                (insert line-breaking))
1519               (t
1520                (insert (format "(%-18s" name))
1521                (setq ret (prin1-to-string value))
1522                (unless (< (+ (current-column)
1523                              (length ret)
1524                              3)
1525                           76)
1526                  (insert line-breaking))
1527                (insert " . " ret ")" line-breaking)
1528                ;; (insert (format "(%-18s . %S)%s"
1529                ;;                 name value
1530                ;;                 line-breaking))
1531                )
1532               ))
1533       (setq attributes (cdr attributes)))
1534     (insert ")")))
1535
1536 (defun insert-char-data (char &optional readable
1537                               attributes)
1538   (save-restriction
1539     (narrow-to-region (point)(point))
1540     (insert "(define-char
1541   '")
1542     (insert-char-attributes char readable attributes)
1543     (insert ")\n")
1544     (goto-char (point-min))
1545     (while (re-search-forward "[ \t]+$" nil t)
1546       (replace-match ""))
1547     ;; from tabify.
1548     (goto-char (point-min))
1549     (while (re-search-forward "[ \t][ \t][ \t]*" nil t)
1550       (let ((column (current-column))
1551             (indent-tabs-mode t))
1552         (delete-region (match-beginning 0) (point))
1553         (indent-to column)))
1554     (goto-char (point-max))
1555     ;; (tabify (point-min)(point-max))
1556     ))
1557
1558 (defun insert-char-data-with-variant (char &optional printable
1559                                            no-ucs-unified
1560                                            script excluded-script)
1561   (insert-char-data char printable)
1562   (let ((variants (char-variants char))
1563         rest
1564         variant vs ret)
1565     (setq variants (sort variants #'<))
1566     (setq rest variants)
1567     (setq variants (cons char variants))
1568     (while rest
1569       (setq variant (car rest))
1570       (unless (get-char-attribute variant '<-subsumptive)
1571         (if (and (or (null script)
1572                      (null (setq vs (get-char-attribute variant 'script)))
1573                      (memq script vs))
1574                  (or (null excluded-script)
1575                      (null (setq vs (get-char-attribute variant 'script)))
1576                      (not (memq excluded-script vs))))
1577             (unless (and no-ucs-unified (get-char-attribute variant '=ucs))
1578               (insert-char-data variant printable)
1579               (if (setq ret (char-variants variant))
1580                   (while ret
1581                     (or (memq (car ret) variants)
1582                         ;; (get-char-attribute (car ret) '<-subsumptive)
1583                         (setq rest (nconc rest (list (car ret)))))
1584                     (setq ret (cdr ret)))))))
1585       (setq rest (cdr rest)))))
1586
1587 (defun insert-char-range-data (min max &optional script excluded-script)
1588   (let ((code min)
1589         char)
1590     (while (<= code max)
1591       (setq char (decode-char '=ucs code))
1592       (if (encode-char char '=ucs 'defined-only)
1593           (insert-char-data-with-variant char nil 'no-ucs-unified
1594                                          script excluded-script))
1595       (setq code (1+ code)))))
1596
1597 (defun write-char-range-data-to-file (min max file
1598                                           &optional script excluded-script)
1599   (let ((coding-system-for-write char-db-file-coding-system))
1600     (with-temp-buffer
1601       (insert (format ";; -*- coding: %s -*-\n"
1602                       char-db-file-coding-system))
1603       (insert-char-range-data min max script excluded-script)
1604       (write-region (point-min)(point-max) file))))
1605
1606 (defvar what-character-original-window-configuration)
1607
1608 ;;;###autoload
1609 (defun what-char-definition (char)
1610   (interactive (list (char-after)))
1611   (let ((buf (get-buffer-create "*Character Description*"))
1612         (the-buf (current-buffer))
1613         (win-conf (current-window-configuration)))
1614     (pop-to-buffer buf)
1615     (make-local-variable 'what-character-original-window-configuration)
1616     (setq what-character-original-window-configuration win-conf)
1617     (setq buffer-read-only nil)
1618     (erase-buffer)
1619     (condition-case err
1620         (progn
1621           (insert-char-data-with-variant char 'printable)
1622           (unless (char-attribute-alist char)
1623             (insert (format ";; = %c\n"
1624                             (let* ((rest (split-char char))
1625                                    (ccs (pop rest))
1626                                    (code (pop rest)))
1627                               (while rest
1628                                 (setq code (logior (lsh code 8)
1629                                                    (pop rest))))
1630                               (decode-char ccs code)))))
1631           ;; (char-db-update-comment)
1632           (set-buffer-modified-p nil)
1633           (view-mode the-buf (lambda (buf)
1634                                (set-window-configuration
1635                                 what-character-original-window-configuration)
1636                                ))
1637           (goto-char (point-min)))
1638       (error (progn
1639                (set-window-configuration
1640                 what-character-original-window-configuration)
1641                (signal (car err) (cdr err)))))))
1642
1643
1644 ;;; @ end
1645 ;;;
1646
1647 (provide 'char-db-util)
1648
1649 ;;; char-db-util.el ends here