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