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