lisp/utf-2000/char-db-util.el

   1 ;;; char-db-util.el --- Character Database utility -*- coding: utf-8-er; -*-
   2
   3 ;; Copyright (C) 1998, 1999, 2000, 2001, 2002, 2003, 2004, 2005, 2006, 2007,
   4 ;;   2008, 2009, 2010, 2011, 2012, 2013, 2014, 2015, 2016, 2017, 2018, 2019,
   5 ;;   2020, 2021, 2022 MORIOKA Tomohiko.
   6
   7 ;; Author: MORIOKA Tomohiko <tomo@kanji.zinbun.kyoto-u.ac.jp>
   8 ;; Keywords: CHISE, Character Database, ISO/IEC 10646, UCS, Unicode, MULE.
   9
  10 ;; This file is part of XEmacs CHISE.
  11
  12 ;; XEmacs CHISE is free software; you can redistribute it and/or
  13 ;; modify it under the terms of the GNU General Public License as
  14 ;; published by the Free Software Foundation; either version 2, or (at
  15 ;; your option) any later version.
  16
  17 ;; XEmacs CHISE is distributed in the hope that it will be useful,
  18 ;; but WITHOUT ANY WARRANTY; without even the implied warranty of
  19 ;; MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the GNU
  20 ;; General Public License for more details.
  21
  22 ;; You should have received a copy of the GNU General Public License
  23 ;; along with XEmacs CHISE; see the file COPYING.  If not, write to
  24 ;; the Free Software Foundation, Inc., 59 Temple Place - Suite 330,
  25 ;; Boston, MA 02111-1307, USA.
  26
  27 ;;; Code:
  28
  29 (require 'chise-subr)
  30 (require 'ideograph-subr)
  31
  32 (defconst unidata-normative-category-alist
  33   '(("Lu" letter        uppercase)
  34     ("Ll" letter        lowercase)
  35     ("Lt" letter        titlecase)
  36     ("Mn" mark          non-spacing)
  37     ("Mc" mark          spacing-combining)
  38     ("Me" mark          enclosing)
  39     ("Nd" number        decimal-digit)
  40     ("Nl" number        letter)
  41     ("No" number        other)
  42     ("Zs" separator     space)
  43     ("Zl" separator     line)
  44     ("Zp" separator     paragraph)
  45     ("Cc" other         control)
  46     ("Cf" other         format)
  47     ("Cs" other         surrogate)
  48     ("Co" other         private-use)
  49     ("Cn" other         not-assigned)))
  50
  51 (defconst unidata-informative-category-alist
  52   '(("Lm" letter        modifier)
  53     ("Lo" letter        other)
  54     ("Pc" punctuation   connector)
  55     ("Pd" punctuation   dash)
  56     ("Ps" punctuation   open)
  57     ("Pe" punctuation   close)
  58     ("Pi" punctuation   initial-quote)
  59     ("Pf" punctuation   final-quote)
  60     ("Po" punctuation   other)
  61     ("Sm" symbol        math)
  62     ("Sc" symbol        currency)
  63     ("Sk" symbol        modifier)
  64     ("So" symbol        other)
  65     ))
  66
  67 (defconst shuowen-radicals
  68   [?一 ?上 ?示 ?三 ?王 ?玉 ?玨 ?气 ?士 ?丨 ; 010
  69    ?屮 ?艸 ?蓐 ?茻 ?小 ?八 ?釆 ?半 ?牛 ?犛 ; 020
  70    ?告 ?口 ?凵 ?吅 ?哭 ?走 ?止 ?癶 ?步 ?此 ; 030
  71    ?正 ?是 ?辵 ?彳 ?廴 ?㢟 ?行 ?齒 ?牙 ?足 ; 040
  72    ?疋 ?品 ?龠 ?冊 ?㗊 ?舌 ?干 ?𧮫 ?只 ?㕯 ; 050
  73    ?句 ?丩 ?古 ?十 ?卅 ?言 ?誩 ?音 ?䇂 ?丵 ; 060
  74    ?菐 ?𠬞 ?𠬜 ?共 ?異 ?舁 ?𦥑 ?䢅 ?爨 ?革 ; 070
  75    ?鬲 ?䰜 ?爪 ?𠃨 ?鬥 ?又 ?𠂇 ?史 ?支 ?𦘒 ; 080
  76    ?聿 ?畫 ?隶 ?臤 ?臣 ?殳 ?殺 ?𠘧 ?寸 ?皮 ; 090
  77    ?㼱 ?攴 ?敎 ?卜 ?用 ?爻 ?㸚 ?𥄎 ?目 ?䀠 ; 100
  78    ?眉 ?盾 ?自 ?白 ?鼻 ?皕 ?習 ?羽 ?隹 ?奞 ; 110
  79    ?雈 ?𦫳 ?𥄕 ?羊 ?羴 ?瞿 ?雔 ?雥 ?鳥 ?烏 ; 120
  80    ?𠦒 ?冓 ?幺 ?𢆶 ?叀 ?玄 ?予 ?放 ?𠬪 ?𣦼 ; 130
  81    ?歺 ?死 ?冎 ?骨 ?肉 ?筋 ?刀 ?刃 ?㓞 ?丯 ; 140
  82    ?耒 ?角 ?竹 ?箕 ?丌 ?左 ?工 ?㠭 ?巫 ?甘 ; 150
  83    ?曰 ?乃 ?丂 ?可 ?兮 ?号 ?亏 ?旨 ?喜 ?壴 ; 160
  84 ;  ?旨 ?曰 ?乃 ?丂 ?可 ?兮 ?号 ?亏 ?喜 ?壴 ; 160
  85    ?鼓 ?豈 ?豆 ?豊 ?豐 ?䖒 ?虍 ?虎 ?虤 ?皿 ; 170
  86    ?𠙴 ?去 ?血 ?丶 ?丹 ?青 ?井 ?皀 ?鬯 ?食 ; 180
  87    ?亼 ?會 ?倉 ?入 ?缶 ?矢 ?高 ?冂 ?𩫏 ?京 ; 190
  88    ?亯 ?𣆪 ?畗 ?㐭 ?嗇 ?來 ?麥 ?夊 ?舛 ?䑞 ; 200
  89    ?韋 ?弟 ?夂 ?久 ?桀 ?木 ?東 ?林 ?才 ?叒 ; 210
  90    ?之 ?帀 ?出 ?𣎵 ?生 ?乇 ?𠂹 ?𠌶 ?華 ?𥝌 ; 220
  91    ?稽 ?巢 ?桼 ?束 ?㯻 ?囗 ?員 ?貝 ?邑 ?𨛜 ; 230
  92    ?日 ?旦 ?倝 ?㫃 ?冥 ?晶 ?月 ?有 ?明 ?囧 ; 240
  93    ?夕 ?多 ?毌 ?𢎘 ?𣐺 ?卣 ?齊 ?朿 ?片 ?鼎 ; 250
  94    ?克 ?彔 ?禾 ?秝 ?黍 ?香 ?米 ?毇 ?臼 ?凶 ; 260
  95    ?𣎳 ?𣏟 ?麻 ?尗 ?耑 ?韭 ?瓜 ?瓠 ?宀 ?宮 ; 270
  96    ?呂 ?穴 ?㝱 ?疒 ?冖 ?𠔼 ?冃 ?㒳 ?网 ?襾 ; 280
  97    ?巾 ?巿 ?帛 ?白 ?㡀 ?黹 ?人 ?𠤎 ?匕 ?从 ; 290
  98    ?比 ?北 ?丘 ?㐺 ?𡈼 ?重 ?臥 ?身 ?㐆 ?衣 ; 300
  99    ?裘 ?老 ?毛 ?毳 ?尸 ?尺 ?尾 ?履 ?舟 ?方 ; 310
 100    ?儿 ?兄 ?兂 ?皃 ?𠑹 ?先 ?秃 ?見 ?覞 ?欠 ; 320
 101    ?㱃 ?㳄 ?旡 ?頁 ?𦣻 ?面 ?丏 ?首 ?𥄉 ?須 ; 330
 102    ?彡 ?彣 ?文 ?髟 ?后 ?司 ?卮 ?卩 ?印 ?色 ; 340
 103    ?𠨍 ?辟 ?勹 ?包 ?茍 ?鬼 ?甶 ?厶 ?嵬 ?山 ; 350
 104    ?屾 ?屵 ?广 ?厂 ?丸 ?危 ?石 ?長 ?勿 ?冄 ; 360
 105    ?而 ?豕 ?㣇 ?彑 ?豚 ?豸 ?𤉡 ?易 ?象 ?馬 ; 370
 106    ?𢊁 ?鹿 ?麤 ?㲋 ?兔 ?萈 ?犬 ?㹜 ?鼠 ?能 ; 380
 107    ?熊 ?火 ?炎 ?黑 ?囪 ?焱 ?炙 ?赤 ?大 ?亦 ; 390
 108    ?夨 ?夭 ?交 ?尣 ?壺 ?壹 ?㚔 ?奢 ?亢 ?夲 ; 400
 109    ?夰 ?亣 ?夫 ?立 ?竝 ?囟 ?思 ?心 ?惢 ?水 ; 410
 110    ?沝 ?瀕 ?𡿨 ?巜 ?川 ?泉 ?灥 ?永 ?𠂢 ?谷 ; 420
 111    ?仌 ?雨 ?雲 ?魚 ?𩺰 ?燕 ?龍 ?飛 ?非 ?卂 ; 430
 112    ?𠃉 ?不 ?至 ?西 ?鹵 ?鹽 ?戶 ?門 ?耳 ?𦣝 ; 440
 113    ?手 ?𠦬 ?女 ?毋 ?民 ?丿 ?𠂆 ?乁 ?氏 ?氐 ; 450
 114    ?戈 ?戉 ?我 ?亅 ?琴 ?𠃊 ?亡 ?匸 ?匚 ?曲 ; 460
 115    ?甾 ?瓦 ?弓 ?弜 ?弦 ?系 ?糸 ?素 ?絲 ?率 ; 470
 116    ?虫 ?䖵 ?蟲 ?風 ?它 ?龜 ?黽 ?卵 ?二 ?土 ; 480
 117    ?垚 ?堇 ?里 ?田 ?畕 ?黃 ?男 ?力 ?劦 ?金 ; 490
 118    ?幵 ?勺 ?几 ?且 ?斤 ?斗 ?矛 ?車 ?𠂤 ?𨸏 ; 500
 119    ?𨺅 ?厽 ?四 ?宁 ?叕 ?亞 ?五 ?六 ?七 ?九 ; 510
 120    ?禸 ?嘼 ?甲 ?乙 ?丙 ?丁 ?戊 ?己 ?巴 ?庚 ; 520
 121    ?辛 ?辡 ?壬 ?癸 ?子 ?了 ?孨 ?𠫓 ?丑 ?寅 ; 530
 122    ?卯 ?辰 ?巳 ?午 ?未 ?申 ?酉 ?酋 ?戌 ?亥 ; 540
 123    ])
 124
 125 (defun shuowen-radical (number)
 126   (aref shuowen-radicals (1- number)))
 127
 128 (defvar char-db-file-coding-system 'utf-8-mcs-er)
 129
 130 (defvar char-db-ignored-attributes '(ideographic-products))
 131
 132 (defvar char-db-coded-charset-priority-list
 133   '(ascii
 134     control-1
 135     latin-iso8859-1
 136     latin-iso8859-2
 137     latin-iso8859-3
 138     latin-iso8859-4
 139     latin-iso8859-9
 140     latin-jisx0201
 141     cyrillic-iso8859-5
 142     greek-iso8859-7
 143     thai-tis620
 144     ;; =mj
 145     ;; =adobe-japan1-0
 146     ;; =adobe-japan1-1
 147     ;; =adobe-japan1-2
 148     ;; =adobe-japan1-3
 149     ;; =adobe-japan1-4
 150     ;; =adobe-japan1-5
 151     ;; =adobe-japan1-6
 152     =jis-x0208
 153     =jis-x0208@1978
 154     =jis-x0208@1983
 155     japanese-jisx0212
 156     chinese-gb2312
 157     =jis-x0208@1990
 158     chinese-cns11643-1
 159     chinese-cns11643-2
 160     chinese-cns11643-3
 161     chinese-cns11643-4
 162     chinese-cns11643-5
 163     chinese-cns11643-6
 164     chinese-cns11643-7
 165     =jis-x0213-1
 166     =jis-x0213-1@2000
 167     =jis-x0213-1@2004
 168     =jis-x0213-2
 169     korean-ksc5601
 170     chinese-isoir165
 171     katakana-jisx0201
 172     hebrew-iso8859-8
 173     chinese-gb12345
 174     latin-viscii
 175     ethiopic-ucs
 176     =big5-cdp
 177     =hanyo-denshi/ja
 178     =hanyo-denshi/jb
 179     =hanyo-denshi/jc
 180     =hanyo-denshi/jd
 181     =hanyo-denshi/ft
 182     =hanyo-denshi/ia
 183     =hanyo-denshi/ib
 184     =hanyo-denshi/hg
 185     =hanyo-denshi/jt
 186     =hanyo-denshi/ks
 187     =hanyo-denshi/tk
 188     ;; ==mj
 189     ;; ==adobe-japan1-0
 190     ;; ==adobe-japan1-1
 191     ;; ==adobe-japan1-2
 192     ;; ==adobe-japan1-3
 193     ;; ==adobe-japan1-4
 194     ;; ==adobe-japan1-5
 195     ;; ==adobe-japan1-6
 196     ==jis-x0208
 197     ==jis-x0213-1
 198     ==jis-x0213-2
 199     ==jis-x0212
 200     ==hanyo-denshi/ja
 201     ==hanyo-denshi/jb
 202     ==hanyo-denshi/jc
 203     ==hanyo-denshi/ft
 204     ==hanyo-denshi/ia
 205     ==hanyo-denshi/ib
 206     ==hanyo-denshi/hg
 207     ==hanyo-denshi/jt
 208     ==hanyo-denshi/ks
 209     ==hanyo-denshi/tk
 210     =gt
 211     =gt-k
 212     =daikanwa
 213     =daijiten
 214     =daikanwa@rev2
 215     =daikanwa@rev1
 216     =daikanwa/+p
 217     =daikanwa/ho
 218     ==daikanwa
 219     ==daijiten
 220     =cbeta
 221     ideograph-hanziku-1
 222     ideograph-hanziku-2
 223     ideograph-hanziku-3
 224     ideograph-hanziku-4
 225     ideograph-hanziku-5
 226     ideograph-hanziku-6
 227     ideograph-hanziku-7
 228     ideograph-hanziku-8
 229     ideograph-hanziku-9
 230     ideograph-hanziku-10
 231     ideograph-hanziku-11
 232     ideograph-hanziku-12
 233     ;; =>>>adobe-japan1-0
 234     ;; =>>>adobe-japan1-1
 235     ;; =>>>adobe-japan1-2
 236     ;; =>>>adobe-japan1-3
 237     ;; =>>>adobe-japan1-4
 238     ;; =>>>adobe-japan1-5
 239     ;; =>>>adobe-japan1-6
 240     ;; =>>>jis-x0208
 241     ;; =>>>jis-x0213-1
 242     ;; =>>>jis-x0213-2
 243     ;; =>>>hanyo-denshi/ja
 244     ;; =>>>hanyo-denshi/jb
 245     ;; =>>>hanyo-denshi/jc
 246     ;; =>>>hanyo-denshi/ft
 247     ;; =>>>hanyo-denshi/ib
 248     ;; =>>>hanyo-denshi/hg
 249     ;; =>>>hanyo-denshi/jt
 250     ;; =>>>hanyo-denshi/ks
 251     ;; =>>>gt
 252     =>>adobe-japan1-0
 253     =>>adobe-japan1-1
 254     =>>adobe-japan1-2
 255     =>>adobe-japan1-3
 256     =>>adobe-japan1-4
 257     =>>adobe-japan1-5
 258     =>>adobe-japan1-6
 259     =>>jis-x0208
 260     =>>jis-x0213-1
 261     =>>jis-x0213-1@2000
 262     =>>jis-x0213-1@2004
 263     =>>jis-x0213-2
 264     =>>jis-x0208@1978
 265     =>>hanyo-denshi/ft
 266     =>>hanyo-denshi/jt
 267     =>>hanyo-denshi/ks
 268     =>>gt
 269     =>>daikanwa
 270     =>>cbeta
 271     =+>jis-x0208
 272     =+>jis-x0213-1
 273     =+>jis-x0213-2
 274     =+>adobe-japan1-0
 275     =+>adobe-japan1-1
 276     =+>adobe-japan1-2
 277     =+>adobe-japan1-3
 278     =+>adobe-japan1-4
 279     =+>adobe-japan1-5
 280     =+>adobe-japan1-6
 281     =+>jis-x0208@1978
 282     =>jis-x0208
 283     =>jis-x0208@1997
 284     =>jis-x0213-1
 285     =>jis-x0213-1@2000
 286     =>jis-x0213-1@2004
 287     =>jis-x0213-2
 288     ==>ucs@bucs
 289     ==>daijiten
 290     ;; =>ucs@hanyo-denshi
 291     =>ucs@iso
 292     =>ucs@unicode
 293     =>ucs@jis
 294     =>ucs@cns
 295     =>ucs@ks
 296     =+>ucs@iso
 297     =+>ucs@unicode
 298     =+>ucs@jis
 299     =+>ucs@jis/1990
 300     =+>ucs@cns
 301     =+>ucs@ks
 302     =>>ucs@iso
 303     =>>ucs@unicode
 304     =>>ucs@jis
 305     =>>ucs@cns
 306     =>>>ucs@iso
 307     =>>>ucs@unicode
 308     ==ucs@iso
 309     ==ucs@unicode
 310     ;; ==ucs@cns
 311     ==gb2312
 312     ==ks-x1001
 313     ==cns11643-1
 314     ==cns11643-2
 315     ==cns11643-3
 316     ==cns11643-4
 317     ==cns11643-5
 318     ==cns11643-6
 319     ==cns11643-7
 320     ==gt
 321     ==jis-x0208@1990
 322     ;; ==jis-x0208@1983
 323     ==jis-x0208@1978
 324     ==gt-k
 325     =ucs@iso
 326     =ucs@unicode
 327     =ucs@cns
 328     ==big5-cdp
 329     ==cbeta
 330     =>>big5-cdp
 331     =>>gt-k
 332     =+>gt
 333     =+>big5-cdp
 334     =>big5-cdp
 335     =>big5-cdp@iwds-1
 336     =>mj
 337     =>hanyo-denshi/tk
 338     =>iwds-1
 339     =big5
 340     =big5-eten
 341     =zinbun-oracle
 342     =>zinbun-oracle
 343     =ruimoku-v6
 344     =>>ruimoku-v6
 345     ==ruimoku-v6
 346     =>ruimoku-v6
 347     =jef-china3
 348     =shinjigen
 349     =>cns11643-5
 350     =>cns11643-7
 351     =>gt
 352     =>gt-k
 353     =>daikanwa
 354     =>daikanwa/ho
 355     =>cbeta
 356     =ucs-var-001
 357     =ucs-var-002
 358     =ucs-var-003
 359     =ucs-var-004
 360     =ucs-var-005
 361     =ucs-var-006
 362     =ucs-var-007
 363     =ucs-var-008
 364     =ucs-var-010
 365     =ucs-itaiji-001
 366     =ucs-itaiji-002
 367     =ucs-itaiji-003
 368     =ucs-itaiji-004
 369     =ucs-itaiji-005
 370     =ucs-itaiji-006
 371     =ucs-itaiji-007
 372     =ucs-itaiji-008
 373     =ucs-itaiji-009
 374     =ucs-itaiji-010
 375     =ucs-itaiji-011
 376     =ucs-itaiji-012
 377     =ucs-itaiji-014
 378     =ucs-itaiji-016
 379     =+>ucs-var-003
 380     =>ucs-itaiji-001
 381     =>ucs-itaiji-002
 382     =>ucs-itaiji-003
 383     =>ucs-itaiji-004
 384     =>ucs-itaiji-005
 385     =>ucs-itaiji-006
 386     =>ucs-itaiji-007
 387     =>ucs-itaiji-008
 388     =>ucs-itaiji-009
 389     =>ucs-itaiji-010
 390     =>ucs-itaiji-011
 391     =>ucs-itaiji-014
 392     =>ucs-itaiji-016
 393     =big5-cdp-var-001
 394     =big5-cdp-var-002
 395     =big5-cdp-var-003
 396     =big5-cdp-var-004
 397     =big5-cdp-var-005
 398     =big5-cdp-var-007
 399     =big5-cdp-var-010
 400     =big5-cdp-itaiji-001
 401     =big5-cdp-itaiji-002
 402     =big5-cdp-itaiji-003
 403     =big5-cdp-itaiji-004
 404     =big5-cdp-itaiji-005
 405     =>big5-cdp-itaiji-001
 406     =>big5-cdp-itaiji-002
 407     =>big5-cdp-itaiji-003
 408     =>big5-cdp-itaiji-004
 409     =>big5-cdp-var-001
 410     =>ucs@iwds-1
 411     =>ucs@cognate
 412     =>ucs@component
 413     =>ucs@iwds-1/normalized
 414     =>ucs-itaiji-001@iwds-1
 415     =>ucs-itaiji-002@iwds-1
 416     =>ucs-itaiji-003@iwds-1
 417     =>ucs-itaiji-006@iwds-1
 418     =>big5-cdp@component
 419     =>big5-cdp@cognate
 420     ==ucs@gb
 421     =ucs@gb
 422     ==ucs-var-002
 423     =ucs@JP/hanazono
 424     =koseki
 425     =shuowen-jiguge
 426     ===shuowen-jiguge5
 427     ===shuowen-jiguge4))
 428
 429
 430 ;;; @ char-db formatters
 431 ;;;
 432
 433 (defun char-db-make-char-spec (char)
 434   (let (ret char-spec)
 435     (cond ((characterp char)
 436            (cond ((and (setq ret (encode-char char '=ucs 'defined-only))
 437                        (not (and (<= #xE000 ret)(<= ret #xF8FF))))
 438                   (setq char-spec (list (cons '=ucs ret)))
 439                   (cond ((setq ret (get-char-attribute char 'name))
 440                          (setq char-spec (cons (cons 'name ret) char-spec))
 441                          )
 442                         ((setq ret (get-char-attribute char 'name*))
 443                          (setq char-spec (cons (cons 'name* ret) char-spec))
 444                          ))
 445                   )
 446                  ((encode-char char '=mj 'defined-only)
 447                   (setq char-spec nil)
 448                   (dolist (ccs (charset-list))
 449                     (if (and (or (eq ccs '=mj)
 450                                  ;; (eq (charset-property ccs 'iso-ir) 177)
 451                                  (string-match "=ucs@" (symbol-name ccs))
 452                                  )
 453                              (setq ccs (charset-name ccs))
 454                              (null (assq ccs char-spec))
 455                              (setq ret (encode-char char ccs 'defined-only)))
 456                         (setq char-spec (cons (cons ccs ret) char-spec))))
 457                   )
 458                  ((encode-char char '==mj 'defined-only)
 459                   (setq char-spec nil)
 460                   (dolist (ccs (charset-list))
 461                     (if (and (or (eq ccs '==mj)
 462                                  ;; (eq (charset-property ccs 'iso-ir) 177)
 463                                  (string-match "=ucs@" (symbol-name ccs))
 464                                  )
 465                              (setq ccs (charset-name ccs))
 466                              (null (assq ccs char-spec))
 467                              (setq ret (encode-char char ccs 'defined-only)))
 468                         (setq char-spec (cons (cons ccs ret) char-spec))))
 469                   )
 470                  ((encode-char char '=adobe-japan1 'defined-only)
 471                   (setq char-spec nil)
 472                   (dolist (ccs (charset-list))
 473                     (if (and (or (memq ccs
 474                                        '(=adobe-japan1-0
 475                                          =adobe-japan1-1
 476                                          =adobe-japan1-2
 477                                          =adobe-japan1-3
 478                                          =adobe-japan1-4
 479                                          =adobe-japan1-5
 480                                          =adobe-japan1-6
 481                                          ))
 482                                  ;; (eq (charset-property ccs 'iso-ir) 177)
 483                                  (string-match "=ucs@" (symbol-name ccs))
 484                                  )
 485                              (setq ccs (charset-name ccs))
 486                              (null (assq ccs char-spec))
 487                              (setq ret (encode-char char ccs 'defined-only)))
 488                         (setq char-spec (cons (cons ccs ret) char-spec))))
 489                   )
 490                  ((encode-char char '==adobe-japan1 'defined-only)
 491                   (setq char-spec nil)
 492                   (dolist (ccs (charset-list))
 493                     (if (and (or (memq ccs
 494                                        '(==adobe-japan1-0
 495                                          ==adobe-japan1-1
 496                                          ==adobe-japan1-2
 497                                          ==adobe-japan1-3
 498                                          ==adobe-japan1-4
 499                                          ==adobe-japan1-5
 500                                          ==adobe-japan1-6
 501                                          ))
 502                                  ;; (eq (charset-property ccs 'iso-ir) 177)
 503                                  (string-match "=ucs@" (symbol-name ccs))
 504                                  )
 505                              (setq ccs (charset-name ccs))
 506                              (null (assq ccs char-spec))
 507                              (setq ret (encode-char char ccs 'defined-only)))
 508                         (setq char-spec (cons (cons ccs ret) char-spec))))
 509                   )
 510                  ((setq ret
 511                         (catch 'tag
 512                           (let ((rest char-db-coded-charset-priority-list)
 513                                 ccs)
 514                             (while rest
 515                               (setq ccs (charset-name
 516                                          (find-charset (car rest))))
 517                               (if (setq ret
 518                                         (encode-char char ccs
 519                                                      'defined-only))
 520                                   (throw 'tag (cons ccs ret)))
 521                               (setq rest (cdr rest))))))
 522                   (setq char-spec (list ret))
 523                   (dolist (ccs (delq (car ret) (charset-list)))
 524                     (if (and (or (charset-iso-final-char ccs)
 525                                  (memq ccs
 526                                        '(=daikanwa
 527                                          =daikanwa@rev2
 528                                          ;; =gt-k
 529                                          =jis-x0208@1997
 530                                          ))
 531                                  (and (eq (charset-property ccs 'iso-ir) 177)
 532                                       (not (eq (charset-name ccs) '=ucs@big5)))
 533                                  ;; (string-match "=ucs@" (symbol-name ccs))
 534                                  )
 535                              (setq ccs (charset-name ccs))
 536                              (null (assq ccs char-spec))
 537                              (setq ret (encode-char char ccs 'defined-only)))
 538                         (setq char-spec (cons (cons ccs ret) char-spec))))
 539                   (if (null char-spec)
 540                       (setq char-spec (split-char char)))
 541                   (cond ((setq ret (get-char-attribute char 'name))
 542                          (setq char-spec (cons (cons 'name ret) char-spec))
 543                          )
 544                         ((setq ret (get-char-attribute char 'name*))
 545                          (setq char-spec (cons (cons 'name* ret) char-spec))
 546                          ))
 547                   )
 548                  ((setq ret (get-char-attribute
 549                              char 'ideographic-combination))
 550                   (setq char-spec
 551                         (cons (cons 'ideographic-combination ret)
 552                               char-spec))
 553                   ))
 554            char-spec)
 555           ((consp char)
 556            char))))
 557
 558 (defun char-db-insert-char-spec (char &optional readable column
 559                                       required-features)
 560   (unless column
 561     (setq column (current-column)))
 562   (let (char-spec temp-char)
 563     (setq char-spec (char-db-make-char-spec char))
 564     (unless (or (characterp char) ; char
 565                 (condition-case nil
 566                     (setq char (find-char char-spec))
 567                   (error nil)))
 568       ;; define temporary character
 569       ;;   Current implementation is dirty.
 570       (setq temp-char (define-char (cons '(ideograph-daikanwa . 0)
 571                                          char-spec)))
 572       (remove-char-attribute temp-char 'ideograph-daikanwa)
 573       (setq char temp-char))
 574     (insert-char-attributes char
 575                             readable
 576                             (union (mapcar #'car char-spec)
 577                                    required-features)
 578                             nil 'for-sub-node)
 579     (when temp-char
 580       ;; undefine temporary character
 581       ;;   Current implementation is dirty.
 582       (setq char-spec (char-attribute-alist temp-char))
 583       (while char-spec
 584         (remove-char-attribute temp-char (car (car char-spec)))
 585         (setq char-spec (cdr char-spec))))))
 586
 587 (defun char-db-insert-alist (alist &optional readable column)
 588   (unless column
 589     (setq column (current-column)))
 590   (let ((line-breaking
 591          (concat "\n" (make-string (1+ column) ?\ )))
 592         name value
 593         ret al ; cal
 594         key
 595         lbs cell rest separator)
 596     (insert "(")
 597     (while alist
 598       (setq name (car (car alist))
 599             value (cdr (car alist)))
 600       (cond ((eq name 'char)
 601              (insert "(char . ")
 602              (if (setq ret (condition-case nil
 603                                (find-char value)
 604                              (error nil)))
 605                  (progn
 606                    (setq al nil
 607                          ;; cal nil
 608                          )
 609                    (while value
 610                      (setq key (car (car value)))
 611                      ;; (if (find-charset key)
 612                      ;;     (setq cal (cons key cal))
 613                      (setq al (cons key al))
 614                      ;; )
 615                      (setq value (cdr value)))
 616                    (insert-char-attributes ret
 617                                            readable
 618                                            (or al 'none) ; cal
 619                                            nil 'for-sub-node))
 620                (insert (prin1-to-string value)))
 621              (insert ")")
 622              (insert line-breaking))
 623             ((consp value)
 624              (insert (format "(%-18s " name))
 625              (setq lbs (concat "\n" (make-string (current-column) ?\ )))
 626              (while (consp value)
 627                (setq cell (car value))
 628                (if (and (consp cell)
 629                         (consp (car cell))
 630                         (setq ret (condition-case nil
 631                                       (find-char cell)
 632                                     (error nil)))
 633                         )
 634                    (progn
 635                      (setq rest cell
 636                            al nil
 637                            ;; cal nil
 638                            )
 639                      (while rest
 640                        (setq key (car (car rest)))
 641                        ;; (if (find-charset key)
 642                        ;;     (setq cal (cons key cal))
 643                        (setq al (cons key al))
 644                        ;; )
 645                        (setq rest (cdr rest)))
 646                      (if separator
 647                          (insert lbs))
 648                      (insert-char-attributes ret
 649                                              readable
 650                                              al ; cal
 651                                              nil 'for-sub-node)
 652                      (setq separator lbs))
 653                  (if separator
 654                      (insert separator))
 655                  (insert (prin1-to-string cell))
 656                  (setq separator " "))
 657                (setq value (cdr value)))
 658              (insert ")")
 659              (insert line-breaking))
 660             (t
 661              (insert (format "(%-18s . %S)%s"
 662                              name value
 663                              line-breaking))))
 664       (setq alist (cdr alist))))
 665   (insert ")"))
 666
 667 (defun char-db-insert-char-reference (plist &optional readable column)
 668   (unless column
 669     (setq column (current-column)))
 670   (let ((line-breaking
 671          (concat "\n" (make-string (1+ column) ?\ )))
 672         (separator "")
 673         name value)
 674     (insert "(")
 675     (while plist
 676       (setq name (pop plist))
 677       (setq value (pop plist))
 678       (cond ((eq name :char)
 679              (insert separator)
 680              (insert ":char\t")
 681              (cond ((numberp value)
 682                     (setq value (decode-char '=ucs value)))
 683                    ;; ((consp value)
 684                    ;;  (setq value (or (find-char value)
 685                    ;;                  value)))
 686                    )
 687              (char-db-insert-char-spec value readable)
 688              (insert line-breaking)
 689              (setq separator ""))
 690             ((eq name :radical)
 691              (insert (format "%s%s\t%d ; %c%s"
 692                              separator
 693                              name value
 694                              (ideographic-radical value)
 695                              line-breaking))
 696              (setq separator ""))
 697             (t
 698              (insert (format "%s%s\t%S" separator name value))
 699              (setq separator line-breaking)))
 700       ))
 701   (insert ")"))
 702
 703 (defun char-db-decode-isolated-char (ccs code-point)
 704   (let (ret)
 705     (setq ret
 706           (cond ((eq ccs 'arabic-iso8859-6)
 707                  (decode-char ccs code-point))
 708                 ;; ((eq ccs '=gt)
 709                 ;;  (decode-builtin-char '==gt code-point))
 710                 ((and (memq ccs '(=gt-pj-1
 711                                   =gt-pj-2
 712                                   =gt-pj-3
 713                                   =gt-pj-4
 714                                   =gt-pj-5
 715                                   =gt-pj-6
 716                                   =gt-pj-7
 717                                   =gt-pj-8
 718                                   =gt-pj-9
 719                                   =gt-pj-10
 720                                   =gt-pj-11))
 721                       (setq ret (decode-char ccs code-point))
 722                       (setq ret (encode-char ret '=gt 'defined-only)))
 723                  (decode-builtin-char '=gt ret))
 724                 (t
 725                  (decode-builtin-char ccs code-point))))
 726     (cond ((null ret)
 727            (or (decode-char ccs code-point)
 728                (define-char (list (cons ccs code-point)))))
 729           ((and (<= 0 (char-int ret))
 730                 (<= (char-int ret) #x1F))
 731            (decode-char '=ucs (+ #x2400 (char-int ret))))
 732           ((= (char-int ret) #x7F)
 733            ?\u2421)
 734           (t ret))))
 735
 736 (defvar char-db-convert-obsolete-format t)
 737
 738 (defun char-db-insert-ccs-feature (name value line-breaking)
 739   (cond
 740    ((integerp value)
 741     (insert
 742      (format
 743       (cond
 744        ((memq name '(=>iwds-1
 745                      ==shinjigen
 746                      =shinjigen
 747                      =shinjigen@1ed ==shinjigen@1ed
 748                      =shinjigen@rev ==shinjigen@rev
 749                      =shinjigen/+p@rev ==shinjigen/+p@rev
 750                      ===daikanwa/ho ==daikanwa/ho
 751                      =daikanwa/ho =>>daikanwa/ho =>daikanwa/ho))
 752         "(%-18s .  %04d)\t; %c")
 753        ((eq name '=shinjigen@1ed/24pr)
 754         "(%-18s . %04d)\t; %c")
 755        ((or
 756          (memq name
 757                '(===daikanwa
 758                  ==daikanwa =daikanwa =>>daikanwa =>daikanwa
 759                  =daikanwa@rev1 =daikanwa@rev2
 760                  =daikanwa/+p ==daikanwa/+p ===daikanwa/+p
 761                  =>>daikanwa/+p
 762                  =daikanwa/+2p =>>daikanwa/+2p
 763                  =gt ==gt ===gt
 764                  =>>gt =+>gt =>gt
 765                  =gt-k ==gt-k ===gt-k
 766                  =>>gt-k =>gt-k
 767                  =adobe-japan1-0 ==adobe-japan1-0 ===adobe-japan1-0
 768                  =adobe-japan1-1 ==adobe-japan1-1 ===adobe-japan1-1
 769                  =adobe-japan1-2 ==adobe-japan1-2 ===adobe-japan1-2
 770                  =adobe-japan1-3 ==adobe-japan1-3 ===adobe-japan1-3
 771                  =adobe-japan1-4 ==adobe-japan1-4 ===adobe-japan1-4
 772                  =adobe-japan1-5 ==adobe-japan1-5 ===adobe-japan1-5
 773                  =adobe-japan1-6 ==adobe-japan1-6 ===adobe-japan1-6
 774                  =>>adobe-japan1-0 =+>adobe-japan1-0
 775                  =>>adobe-japan1-1 =+>adobe-japan1-1
 776                  =>>adobe-japan1-2 =+>adobe-japan1-2
 777                  =>>adobe-japan1-3 =+>adobe-japan1-3
 778                  =>>adobe-japan1-4 =+>adobe-japan1-4
 779                  =>>adobe-japan1-5 =+>adobe-japan1-5
 780                  =>>adobe-japan1-6 =+>adobe-japan1-6
 781                  =>cbeta =cbeta =>>cbeta ==cbeta ===cbeta
 782                  =zinbun-oracle =>zinbun-oracle
 783                  =daijiten ==daijiten ===daijiten ==>daijiten
 784                  ===hng-jou ===hng-keg ===hng-dng ===hng-mam
 785                  ===hng-drt ===hng-kgk ===hng-myz ===hng-kda
 786                  ===hng-khi ===hng-khm ===hng-fhs ===hng-hok
 787                  ===hng-kyd ===hng-sok
 788                  ===hng-yhk ===hng-kak ===hng-kar ===hng-kae
 789                  ===hng-sys ===hng-tsu ===hng-tzj
 790                  ===hng-hos ===hng-kkh ===hng-nak ===hng-jhk
 791                  ===hng-hod ===hng-gok ===hng-ink ===hng-nto
 792                  ===hng-nkm ===hng-k24 ===hng-ini ===hng-nkk
 793                  ===hng-kcc ===hng-kcj ===hng-kbk ===hng-sik
 794                  ===hng-skk ===hng-kyu ===hng-ksk ===hng-wan
 795                  ===hng-okd ===hng-wad ===hng-kmi
 796                  ===hng-zkd ===hng-doh ===hng-jyu ===hng-tzs
 797                  ===hng-sai ===hng-kad ===hng-kss
 798                  ===hng-kyo ===hng-ykk ===hng-saa
 799                  ===hng-sab ===hng-wks ===hng-wke ===hng-smk
 800                  ===hng-sgs ===hng-sts ===hng-kkd
 801                  =shuowen-jiguge ===shuowen-jiguge4 ===shuowen-jiguge5))
 802          ;; (string-match "^=adobe-" (symbol-name name))
 803          )
 804         "(%-18s . %05d)\t; %c")
 805        ((memq name '(=hanyo-denshi/ks
 806                      ==hanyo-denshi/ks ===hanyo-denshi/ks
 807                      =>>hanyo-denshi/ks
 808                      =koseki ==koseki
 809                      =mj ==mj ===mj =>>mj =>mj
 810                      =zihai mojikyo))
 811         "(%-18s . %06d)\t; %c")
 812        ((memq name '(=hanyo-denshi/tk =>hanyo-denshi/tk ==hanyo-denshi/tk))
 813         "(%-18s . %08d)\t; %c")
 814        ((>= (charset-dimension name) 2)
 815         "(%-18s . #x%04X)\t; %c")
 816        (t
 817         "(%-18s . #x%02X)\t; %c"))
 818       name
 819       (if (= (charset-iso-graphic-plane name) 1)
 820           (logior value
 821                   (cond ((= (charset-dimension name) 1)
 822                          #x80)
 823                         ((= (charset-dimension name) 2)
 824                          #x8080)
 825                         ((= (charset-dimension name) 3)
 826                          #x808080)
 827                         (t 0)))
 828         value)
 829       (char-db-decode-isolated-char name value)))
 830     (if (and (= (charset-chars name) 94)
 831              (= (charset-dimension name) 2))
 832         (insert (format " [%02d-%02d]"
 833                         (- (lsh value -8) 32)
 834                         (- (logand value 255) 32))))
 835     )
 836    (t
 837     (insert (format "(%-18s . %s)" name value))
 838     ))
 839   (insert line-breaking))
 840
 841 (defun char-db-insert-relation-feature (char name value line-breaking
 842                                              ccss readable)
 843   (insert (format "(%-18s%s " name line-breaking))
 844   (let ((lbs (concat "\n" (make-string (current-column) ?\ )))
 845         separator cell sources required-features
 846         ret)
 847     (while (consp value)
 848       (setq cell (car value))
 849       (if (integerp cell)
 850           (setq cell (decode-char '=ucs cell)))
 851       (cond
 852        ((eq name '->subsumptive)
 853         (when (or (not (some (lambda (atr)
 854                                (get-char-attribute cell atr))
 855                              char-db-ignored-attributes))
 856                   (some (lambda (ccs)
 857                           (encode-char cell ccs 'defined-only))
 858                         ccss))
 859           (if separator
 860               (insert lbs))
 861           (let ((char-db-ignored-attributes
 862                  (cons '<-subsumptive
 863                        char-db-ignored-attributes)))
 864             (insert-char-attributes cell readable nil nil 'for-sub-node))
 865           (setq separator lbs))
 866         )
 867        ((characterp cell)
 868         (setq sources
 869               (get-char-attribute
 870                char (intern (format "%s*sources" name))))
 871         (setq required-features nil)
 872         (dolist (source sources)
 873           (cond
 874            ((memq source '(JP
 875                            JP/Jouyou
 876                            shinjigen shinjigen@1ed shinjigen@rev))
 877             (setq required-features
 878                   (union required-features
 879                          '(=jis-x0208
 880                            =jis-x0208@1990
 881                            =jis-x0213-1@2000
 882                            =jis-x0213-1@2004
 883                            =jis-x0213-2
 884                            =jis-x0212
 885                            =jis-x0208@1983
 886                            =jis-x0208@1978
 887                            =shinjigen
 888                            =shinjigen@1ed
 889                            =shinjigen@rev
 890                            =shinjigen/+p@rev))))
 891            ((eq source 'CN)
 892             (setq required-features
 893                   (union required-features
 894                          '(=gb2312
 895                            =gb12345
 896                            =iso-ir165)))))
 897           (cond
 898            ((find-charset (setq ret (intern (format "=%s" source))))
 899             (setq required-features
 900                   (cons ret required-features)))
 901            (t (setq required-features
 902                     (cons source required-features)))))
 903         (cond ((string-match "@JP" (symbol-name name))
 904                (setq required-features
 905                      (union required-features
 906                             '(=jis-x0208
 907                               =jis-x0208@1990
 908                               =jis-x0213-1-2000
 909                               =jis-x0213-2-2000
 910                               =jis-x0212
 911                               =jis-x0208@1983
 912                               =jis-x0208@1978))))
 913               ((string-match "@CN" (symbol-name name))
 914                (setq required-features
 915                      (union required-features
 916                             '(=gb2312
 917                               =gb12345
 918                               =iso-ir165)))))
 919         (if separator
 920             (insert lbs))
 921         (if readable
 922             (insert (format "%S" cell))
 923           (char-db-insert-char-spec cell readable
 924                                     nil
 925                                     required-features))
 926         (setq separator lbs))
 927        ((consp cell)
 928         (if separator
 929             (insert lbs))
 930         (if (consp (car cell))
 931             (char-db-insert-char-spec cell readable)
 932           (char-db-insert-char-reference cell readable))
 933         (setq separator lbs))
 934        (t
 935         (if separator
 936             (insert separator))
 937         (insert (prin1-to-string cell))
 938         (setq separator " ")))
 939       (setq value (cdr value)))
 940     (insert ")")
 941     (insert line-breaking)))
 942
 943 (defun insert-char-attributes (char &optional readable attributes column
 944                                     for-sub-node)
 945   (unless column
 946     (setq column (current-column)))
 947   (let (name value ; has-long-ccs-name
 948         rest
 949         radical strokes
 950         (line-breaking
 951          (concat "\n" (make-string (1+ column) ?\ )))
 952         lbs cell separator ret
 953         key al cal
 954         dest-ccss ; sources required-features
 955         ccss)
 956     (let (atr-d)
 957       (setq attributes
 958             (sort (if attributes
 959                       (if (consp attributes)
 960                           (progn
 961                             (dolist (name attributes)
 962                               (unless (memq name char-db-ignored-attributes)
 963                                 (if (find-charset name)
 964                                     (push name ccss))
 965                                 (push name atr-d)))
 966                             atr-d))
 967                     (dolist (name (char-attribute-list))
 968                       (unless (memq name char-db-ignored-attributes)
 969                         (if (find-charset name)
 970                             (push name ccss))
 971                         (push name atr-d)))
 972                     atr-d)
 973                   #'char-attribute-name<)))
 974     (insert "(")
 975     (when (memq '<-subsumptive attributes)
 976       (when (or readable (not for-sub-node))
 977         (when (setq value (get-char-attribute char '<-subsumptive))
 978           (char-db-insert-relation-feature char '<-subsumptive value
 979                                            line-breaking
 980                                            ccss readable)))
 981       (setq attributes (delq '<-subsumptive attributes)))
 982     (when (and (memq '<-denotational attributes)
 983                (setq value (get-char-attribute char '<-denotational)))
 984       (char-db-insert-relation-feature char '<-denotational value
 985                                        line-breaking
 986                                        ccss readable)
 987       (setq attributes (delq '<-denotational attributes)))
 988     (when (and (memq '<-denotational@component attributes)
 989                (setq value (get-char-attribute char '<-denotational@component)))
 990       (char-db-insert-relation-feature char '<-denotational@component value
 991                                        line-breaking
 992                                        ccss readable)
 993       (setq attributes (delq '<-denotational@component attributes)))
 994     (when (and (memq '<-denotational@usage attributes)
 995                (setq value (get-char-attribute char '<-denotational@usage)))
 996       (char-db-insert-relation-feature char '<-denotational@usage value
 997                                        line-breaking
 998                                        ccss readable)
 999       (setq attributes (delq '<-denotational@usage attributes)))
1000     (when (and (memq 'name attributes)
1001                (setq value (get-char-attribute char 'name)))
1002       (insert (format
1003                (if (> (+ (current-column) (length value)) 48)
1004                    "(name . %S)%s"
1005                  "(name               . %S)%s")
1006                value line-breaking))
1007       (setq attributes (delq 'name attributes))
1008       )
1009     (when (and (memq 'name* attributes)
1010                (setq value (get-char-attribute char 'name*)))
1011       (insert (format
1012                (if (> (+ (current-column) (length value)) 48)
1013                    "(name* . %S)%s"
1014                  "(name*              . %S)%s")
1015                value line-breaking))
1016       (setq attributes (delq 'name* attributes))
1017       )
1018     (when (and (memq 'script attributes)
1019                (setq value (get-char-attribute char 'script)))
1020       (insert (format "(script\t\t%s)%s"
1021                       (mapconcat (function prin1-to-string)
1022                                  value " ")
1023                       line-breaking))
1024       (setq attributes (delq 'script attributes))
1025       )
1026     (dolist (name '(=>ucs =>ucs*))
1027       (when (and (memq name attributes)
1028                  (setq value (get-char-attribute char name)))
1029         (insert (format "(%-18s . #x%04X)\t; %c%s"
1030                         name value (decode-char '=ucs value)
1031                         line-breaking))
1032         (setq attributes (delq name attributes))))
1033     (dolist (name '(=>ucs@gb =>ucs@big5))
1034       (when (and (memq name attributes)
1035                  (setq value (get-char-attribute char name)))
1036         (insert (format "(%-18s . #x%04X)\t; %c%s"
1037                         name value
1038                         (decode-char (intern
1039                                       (concat "="
1040                                               (substring
1041                                                (symbol-name name) 2)))
1042                                      value)
1043                         line-breaking))
1044         (setq attributes (delq name attributes))
1045         ))
1046     ;; (dolist (name '(=>daikanwa))
1047     ;;   (when (and (memq name attributes)
1048     ;;              (setq value (get-char-attribute char name)))
1049     ;;     (insert
1050     ;;      (if (integerp value)
1051     ;;          (format "(%-18s . %05d)\t; %c%s"
1052     ;;                  name value (decode-char '=daikanwa value)
1053     ;;                  line-breaking)
1054     ;;        (format "(%-18s %s)\t; %c%s"
1055     ;;                name
1056     ;;                (mapconcat (function prin1-to-string)
1057     ;;                           value " ")
1058     ;;                (char-representative-of-daikanwa char)
1059     ;;                line-breaking)))
1060     ;;     (setq attributes (delq name attributes))))
1061     (when (and (memq 'general-category attributes)
1062                (setq value (get-char-attribute char 'general-category)))
1063       (insert (format
1064                "(general-category\t%s) ; %s%s"
1065                (mapconcat (lambda (cell)
1066                             (format "%S" cell))
1067                           value " ")
1068                (cond ((rassoc value unidata-normative-category-alist)
1069                       "Normative Category")
1070                      ((rassoc value unidata-informative-category-alist)
1071                       "Informative Category")
1072                      (t
1073                       "Unknown Category"))
1074                line-breaking))
1075       (setq attributes (delq 'general-category attributes))
1076       )
1077     (when (and (memq 'bidi-category attributes)
1078                (setq value (get-char-attribute char 'bidi-category)))
1079       (insert (format "(bidi-category\t. %S)%s"
1080                       value
1081                       line-breaking))
1082       (setq attributes (delq 'bidi-category attributes))
1083       )
1084     (unless (or (not (memq 'mirrored attributes))
1085                 (eq (setq value (get-char-attribute char 'mirrored 'empty))
1086                     'empty))
1087       (insert (format "(mirrored\t\t. %S)%s"
1088                       value
1089                       line-breaking))
1090       (setq attributes (delq 'mirrored attributes))
1091       )
1092     (cond
1093      ((and (memq 'decimal-digit-value attributes)
1094            (setq value (get-char-attribute char 'decimal-digit-value)))
1095       (insert (format "(decimal-digit-value . %S)%s"
1096                       value
1097                       line-breaking))
1098       (setq attributes (delq 'decimal-digit-value attributes))
1099       (when (and (memq 'digit-value attributes)
1100                  (setq value (get-char-attribute char 'digit-value)))
1101         (insert (format "(digit-value\t . %S)%s"
1102                         value
1103                         line-breaking))
1104         (setq attributes (delq 'digit-value attributes))
1105         )
1106       (when (and (memq 'numeric-value attributes)
1107                  (setq value (get-char-attribute char 'numeric-value)))
1108         (insert (format "(numeric-value\t . %S)%s"
1109                         value
1110                         line-breaking))
1111         (setq attributes (delq 'numeric-value attributes))
1112         )
1113       )
1114      (t
1115       (when (and (memq 'digit-value attributes)
1116                  (setq value (get-char-attribute char 'digit-value)))
1117         (insert (format "(digit-value\t. %S)%s"
1118                         value
1119                         line-breaking))
1120         (setq attributes (delq 'digit-value attributes))
1121         )
1122       (when (and (memq 'numeric-value attributes)
1123                  (setq value (get-char-attribute char 'numeric-value)))
1124         (insert (format "(numeric-value\t. %S)%s"
1125                         value
1126                         line-breaking))
1127         (setq attributes (delq 'numeric-value attributes))
1128         )))
1129     (when (and (memq 'iso-10646-comment attributes)
1130                (setq value (get-char-attribute char 'iso-10646-comment)))
1131       (insert (format "(iso-10646-comment\t. %S)%s"
1132                       value
1133                       line-breaking))
1134       (setq attributes (delq 'iso-10646-comment attributes))
1135       )
1136     (when (and (memq 'morohashi-daikanwa attributes)
1137                (setq value (get-char-attribute char 'morohashi-daikanwa)))
1138       (insert (format "(morohashi-daikanwa\t%s)%s"
1139                       (mapconcat (function prin1-to-string) value " ")
1140                       line-breaking))
1141       (setq attributes (delq 'morohashi-daikanwa attributes))
1142       )
1143     (setq radical nil
1144           strokes nil)
1145     (when (and (memq 'ideographic-radical attributes)
1146                (setq value (get-char-attribute char 'ideographic-radical)))
1147       (setq radical value)
1148       (insert (format "(ideographic-radical . %S)\t; %c%s"
1149                       radical
1150                       (ideographic-radical radical)
1151                       line-breaking))
1152       (setq attributes (delq 'ideographic-radical attributes))
1153       )
1154     (when (and (memq 'shuowen-radical attributes)
1155                (setq value (get-char-attribute char 'shuowen-radical)))
1156       (insert (format "(shuowen-radical\t. %S)\t; %c%s"
1157                       value
1158                       (shuowen-radical value)
1159                       line-breaking))
1160       (setq attributes (delq 'shuowen-radical attributes))
1161       )
1162     (let (key)
1163       (dolist (domain
1164                (append
1165                 char-db-feature-domains
1166                 (let (dest domain)
1167                   (dolist (feature (char-attribute-list))
1168                     (setq feature (symbol-name feature))
1169                     (when (string-match
1170                            "\\(radical\\|strokes\\)@\\([^@*]+\\)\\(\\*\\|$\\)"
1171                            feature)
1172                       (setq domain (intern (match-string 2 feature)))
1173                      (unless (memq domain dest)
1174                        (setq dest (cons domain dest)))))
1175                   (sort dest #'string<))))
1176         (setq key (intern (format "%s@%s" 'ideographic-radical domain)))
1177         (when (and (memq key attributes)
1178                    (setq value (get-char-attribute char key)))
1179           (setq radical value)
1180           (insert (format "(%s . %S)\t; %c%s"
1181                           key
1182                           radical
1183                           (ideographic-radical radical)
1184                           line-breaking))
1185           (setq attributes (delq key attributes))
1186           )
1187         (setq key (intern (format "%s@%s" 'ideographic-strokes domain)))
1188         (when (and (memq key attributes)
1189                    (setq value (get-char-attribute char key)))
1190           (setq strokes value)
1191           (insert (format "(%s . %S)%s"
1192                           key
1193                           strokes
1194                           line-breaking))
1195           (setq attributes (delq key attributes))
1196           )
1197         (setq key (intern (format "%s@%s" 'total-strokes domain)))
1198         (when (and (memq key attributes)
1199                    (setq value (get-char-attribute char key)))
1200           (insert (format "(%s       . %S)%s"
1201                           key
1202                           value
1203                           line-breaking))
1204           (setq attributes (delq key attributes))
1205           )
1206         (dolist (feature '(ideographic-radical
1207                            ideographic-strokes
1208                            total-strokes))
1209           (setq key (intern (format "%s@%s*sources" feature domain)))
1210           (when (and (memq key attributes)
1211                      (setq value (get-char-attribute char key)))
1212             (insert (format "(%s%s" key line-breaking))
1213             (dolist (cell value)
1214               (insert (format " %s" cell)))
1215             (insert ")")
1216             (insert line-breaking)
1217             (setq attributes (delq key attributes))
1218             ))
1219         ))
1220     (when (and (memq 'ideographic-strokes attributes)
1221                (setq value (get-char-attribute char 'ideographic-strokes)))
1222       (setq strokes value)
1223       (insert (format "(ideographic-strokes . %S)%s"
1224                       strokes
1225                       line-breaking))
1226       (setq attributes (delq 'ideographic-strokes attributes))
1227       )
1228     (when (and (memq 'kangxi-radical attributes)
1229                (setq value (get-char-attribute char 'kangxi-radical)))
1230       (unless (eq value radical)
1231         (insert (format "(kangxi-radical\t . %S)\t; %c%s"
1232                         value
1233                         (ideographic-radical value)
1234                         line-breaking))
1235         (or radical
1236             (setq radical value)))
1237       (setq attributes (delq 'kangxi-radical attributes))
1238       )
1239     (when (and (memq 'kangxi-strokes attributes)
1240                (setq value (get-char-attribute char 'kangxi-strokes)))
1241       (unless (eq value strokes)
1242         (insert (format "(kangxi-strokes\t . %S)%s"
1243                         value
1244                         line-breaking))
1245         (or strokes
1246             (setq strokes value)))
1247       (setq attributes (delq 'kangxi-strokes attributes))
1248       )
1249     ;; (when (and (memq 'japanese-radical attributes)
1250     ;;            (setq value (get-char-attribute char 'japanese-radical)))
1251     ;;   (unless (eq value radical)
1252     ;;     (insert (format "(japanese-radical\t . %S)\t; %c%s"
1253     ;;                     value
1254     ;;                     (ideographic-radical value)
1255     ;;                     line-breaking))
1256     ;;     (or radical
1257     ;;         (setq radical value)))
1258     ;;   (setq attributes (delq 'japanese-radical attributes))
1259     ;;   )
1260     (when (and (memq 'japanese-strokes attributes)
1261                (setq value (get-char-attribute char 'japanese-strokes)))
1262       (unless (eq value strokes)
1263         (insert (format "(japanese-strokes\t . %S)%s"
1264                         value
1265                         line-breaking))
1266         (or strokes
1267             (setq strokes value)))
1268       (setq attributes (delq 'japanese-strokes attributes))
1269       )
1270     (when (and (memq 'cns-radical attributes)
1271                (setq value (get-char-attribute char 'cns-radical)))
1272       (insert (format "(cns-radical\t . %S)\t; %c%s"
1273                       value
1274                       (ideographic-radical value)
1275                       line-breaking))
1276       (setq attributes (delq 'cns-radical attributes))
1277       )
1278     (when (and (memq 'cns-strokes attributes)
1279                (setq value (get-char-attribute char 'cns-strokes)))
1280       (unless (eq value strokes)
1281         (insert (format "(cns-strokes\t . %S)%s"
1282                         value
1283                         line-breaking))
1284         (or strokes
1285             (setq strokes value)))
1286       (setq attributes (delq 'cns-strokes attributes))
1287       )
1288     ;; (when (and (memq 'ideographic- attributes)
1289     ;;            (setq value (get-char-attribute char 'ideographic-)))
1290     ;;   (insert "(ideographic-       ")
1291     ;;   (setq lbs (concat "\n" (make-string (current-column) ?\ ))
1292     ;;         separator nil)
1293     ;;   (while (consp value)
1294     ;;     (setq cell (car value))
1295     ;;     (if (integerp cell)
1296     ;;         (setq cell (decode-char '=ucs cell)))
1297     ;;     (cond ((characterp cell)
1298     ;;            (if separator
1299     ;;                (insert lbs))
1300     ;;            (if readable
1301     ;;                (insert (format "%S" cell))
1302     ;;              (char-db-insert-char-spec cell readable))
1303     ;;            (setq separator lbs))
1304     ;;           ((consp cell)
1305     ;;            (if separator
1306     ;;                (insert lbs))
1307     ;;            (if (consp (car cell))
1308     ;;                (char-db-insert-char-spec cell readable)
1309     ;;              (char-db-insert-char-reference cell readable))
1310     ;;            (setq separator lbs))
1311     ;;           (t
1312     ;;            (if separator
1313     ;;                (insert separator))
1314     ;;            (insert (prin1-to-string cell))
1315     ;;            (setq separator " ")))
1316     ;;     (setq value (cdr value)))
1317     ;;   (insert ")")
1318     ;;   (insert line-breaking)
1319     ;;   (setq attributes (delq 'ideographic- attributes)))
1320     (when (and (memq 'total-strokes attributes)
1321                (setq value (get-char-attribute char 'total-strokes)))
1322       (insert (format "(total-strokes       . %S)%s"
1323                       value
1324                       line-breaking))
1325       (setq attributes (delq 'total-strokes attributes))
1326       )
1327     ;; (when (and (memq '->ideograph attributes)
1328     ;;            (setq value (get-char-attribute char '->ideograph)))
1329     ;;   (insert (format "(->ideograph\t%s)%s"
1330     ;;                   (mapconcat (lambda (code)
1331     ;;                                (cond ((symbolp code)
1332     ;;                                       (symbol-name code))
1333     ;;                                      ((integerp code)
1334     ;;                                       (format "#x%04X" code))
1335     ;;                                      (t
1336     ;;                                       (format "%s %S"
1337     ;;                                               line-breaking code))))
1338     ;;                              value " ")
1339     ;;                   line-breaking))
1340     ;;   (setq attributes (delq '->ideograph attributes))
1341     ;;   )
1342     ;; (when (and (memq '->decomposition attributes)
1343     ;;            (setq value (get-char-attribute char '->decomposition)))
1344     ;;   (insert (format "(->decomposition\t%s)%s"
1345     ;;                   (mapconcat (lambda (code)
1346     ;;                                (cond ((symbolp code)
1347     ;;                                       (symbol-name code))
1348     ;;                                      ((characterp code)
1349     ;;                                       (if readable
1350     ;;                                           (format "%S" code)
1351     ;;                                         (format "#x%04X"
1352     ;;                                                 (char-int code))
1353     ;;                                         ))
1354     ;;                                      ((integerp code)
1355     ;;                                       (format "#x%04X" code))
1356     ;;                                      (t
1357     ;;                                       (format "%s%S" line-breaking code))))
1358     ;;                              value " ")
1359     ;;                   line-breaking))
1360     ;;   (setq attributes (delq '->decomposition attributes))
1361     ;;   )
1362     (if (equal (get-char-attribute char '->titlecase)
1363                (get-char-attribute char '->uppercase))
1364         (setq attributes (delq '->titlecase attributes)))
1365     (when (and (memq '->mojikyo attributes)
1366                (setq value (get-char-attribute char '->mojikyo)))
1367       (insert (format "(->mojikyo\t\t. %06d)\t; %c%s"
1368                       value (decode-char 'mojikyo value)
1369                       line-breaking))
1370       (setq attributes (delq '->mojikyo attributes))
1371       )
1372     (unless readable
1373       (dolist (ignored '(composition
1374                          ->denotational <-subsumptive ->ucs-unified
1375                          ;; ->ideographic-component-forms
1376                          ))
1377         (setq attributes (delq ignored attributes))))
1378     (while attributes
1379       (setq name (car attributes))
1380       (unless (eq (setq value (get-char-attribute char name 'value-is-empty))
1381                   'value-is-empty)
1382         (cond ((setq ret (find-charset name))
1383                (setq name (charset-name ret))
1384                (when (not (memq name dest-ccss))
1385                  (setq dest-ccss (cons name dest-ccss))
1386                  (char-db-insert-ccs-feature name value line-breaking))
1387                )
1388               ((string-match "^=>ucs@" (symbol-name name))
1389                (insert (format "(%-18s . #x%04X)\t; %c%s"
1390                                name value (decode-char '=ucs value)
1391                                line-breaking))
1392                )
1393               ((eq name 'jisx0208-1978/4X)
1394                (insert (format "(%-18s . #x%04X)%s"
1395                                name value
1396                                line-breaking))
1397                )
1398               ((and
1399                 (not readable)
1400                 (not (eq name '->subsumptive))
1401                 (not (eq name '->uppercase))
1402                 (not (eq name '->lowercase))
1403                 (not (eq name '->titlecase))
1404                 (not (eq name '->canonical))
1405                 (not (eq name '->Bopomofo))
1406                 (not (eq name '->mistakable))
1407                 (not (eq name '->ideographic-variants))
1408                 (null (get-char-attribute
1409                        char (intern (format "%s*sources" name))))
1410                 (not (string-match "\\*sources$" (symbol-name name)))
1411                 (null (get-char-attribute
1412                        char (intern (format "%s*note" name))))
1413                 (not (string-match "\\*note$" (symbol-name name)))
1414                 (or (eq name '<-identical)
1415                     (eq name '<-uppercase)
1416                     (eq name '<-lowercase)
1417                     (eq name '<-titlecase)
1418                     (eq name '<-canonical)
1419                     (eq name '<-ideographic-variants)
1420                     ;; (eq name '<-synonyms)
1421                     (string-match "^<-synonyms" (symbol-name name))
1422                     (eq name '<-mistakable)
1423                     (when (string-match "^->" (symbol-name name))
1424                       (cond
1425                        ((string-match "^->fullwidth" (symbol-name name))
1426                         (not (and (consp value)
1427                                   (characterp (car value))
1428                                   (encode-char
1429                                    (car value) '=ucs 'defined-only)))
1430                         )
1431                        (t)))
1432                     ))
1433                )
1434               ((or (eq name 'ideographic-structure)
1435                    (eq name 'ideographic-structure@apparent)
1436                    (eq name 'ideographic-structure@apparent/leftmost)
1437                    (eq name 'ideographic-combination)
1438                    ;; (eq name 'ideographic-)
1439                    (eq name '=decomposition)
1440                    (char-feature-base-name= '=decomposition name)
1441                    (char-feature-base-name= '=>decomposition name)
1442                    ;; (string-match "^=>*decomposition\\(@[^*]+\\)?$"
1443                    ;;               (symbol-name name))
1444                    (string-match "^\\(->\\|<-\\)[^*]*$" (symbol-name name))
1445                    (string-match "^\\(->\\|<-\\)[^*]*\\*sources$"
1446                                  (symbol-name name))
1447                    )
1448                (char-db-insert-relation-feature char name value
1449                                                 line-breaking
1450                                                 ccss readable))
1451               ;; ((memq name '(ideograph=
1452               ;;               original-ideograph-of
1453               ;;               ancient-ideograph-of
1454               ;;               vulgar-ideograph-of
1455               ;;               wrong-ideograph-of
1456               ;;               ;; simplified-ideograph-of
1457               ;;               ideographic-variants
1458               ;;               ;; ideographic-different-form-of
1459               ;;               ))
1460               ;;  (insert (format "(%-18s%s " name line-breaking))
1461               ;;  (setq lbs (concat "\n" (make-string (current-column) ?\ ))
1462               ;;        separator nil)
1463               ;;  (while (consp value)
1464               ;;    (setq cell (car value))
1465               ;;    (if (and (consp cell)
1466               ;;             (consp (car cell)))
1467               ;;        (progn
1468               ;;          (if separator
1469               ;;              (insert lbs))
1470               ;;          (char-db-insert-alist cell readable)
1471               ;;          (setq separator lbs))
1472               ;;      (if separator
1473               ;;          (insert separator))
1474               ;;      (insert (prin1-to-string cell))
1475               ;;      (setq separator " "))
1476               ;;    (setq value (cdr value)))
1477               ;;  (insert ")")
1478               ;;  (insert line-breaking))
1479               ((consp value)
1480                (insert (format "(%-18s " name))
1481                (setq lbs (concat "\n" (make-string (current-column) ?\ ))
1482                      separator nil)
1483                (while (consp value)
1484                  (setq cell (car value))
1485                  (if (and (consp cell)
1486                           (consp (car cell))
1487                           (setq ret (condition-case nil
1488                                         (find-char cell)
1489                                       (error nil))))
1490                      (progn
1491                        (setq rest cell
1492                              al nil
1493                              cal nil)
1494                        (while rest
1495                          (setq key (car (car rest)))
1496                          (if (find-charset key)
1497                              (setq cal (cons key cal))
1498                            (setq al (cons key al)))
1499                          (setq rest (cdr rest)))
1500                        (if separator
1501                            (insert lbs))
1502                        (insert-char-attributes ret
1503                                                readable
1504                                                al ; cal
1505                                                nil 'for-sub-node)
1506                        (setq separator lbs))
1507                    (setq ret (prin1-to-string cell))
1508                    (if separator
1509                        (if (< (+ (current-column)
1510                                  (length ret)
1511                                  (length separator))
1512                               76)
1513                            (insert separator)
1514                          (insert lbs)))
1515                    (insert ret)
1516                    (setq separator " "))
1517                  (setq value (cdr value)))
1518                (insert ")")
1519                (insert line-breaking))
1520               (t
1521                (insert (format "(%-18s" name))
1522                (setq ret (prin1-to-string value))
1523                (unless (< (+ (current-column)
1524                              (length ret)
1525                              3)
1526                           76)
1527                  (insert line-breaking))
1528                (insert " . " ret ")" line-breaking)
1529                ;; (insert (format "(%-18s . %S)%s"
1530                ;;                 name value
1531                ;;                 line-breaking))
1532                )
1533               ))
1534       (setq attributes (cdr attributes)))
1535     (insert ")")))
1536
1537 (defun insert-char-data (char &optional readable
1538                               attributes)
1539   (save-restriction
1540     (narrow-to-region (point)(point))
1541     (insert "(define-char
1542   '")
1543     (insert-char-attributes char readable attributes)
1544     (insert ")\n")
1545     (goto-char (point-min))
1546     (while (re-search-forward "[ \t]+$" nil t)
1547       (replace-match ""))
1548     ;; from tabify.
1549     (goto-char (point-min))
1550     (while (re-search-forward "[ \t][ \t][ \t]*" nil t)
1551       (let ((column (current-column))
1552             (indent-tabs-mode t))
1553         (delete-region (match-beginning 0) (point))
1554         (indent-to column)))
1555     (goto-char (point-max))
1556     ;; (tabify (point-min)(point-max))
1557     ))
1558
1559 (defun insert-char-data-with-variant (char &optional printable
1560                                            no-ucs-unified
1561                                            script excluded-script)
1562   (insert-char-data char printable)
1563   (let ((variants (char-variants char))
1564         rest
1565         variant vs ret)
1566     (setq variants (sort variants #'<))
1567     (setq rest variants)
1568     (setq variants (cons char variants))
1569     (while rest
1570       (setq variant (car rest))
1571       (unless (get-char-attribute variant '<-subsumptive)
1572         (if (and (or (null script)
1573                      (null (setq vs (get-char-attribute variant 'script)))
1574                      (memq script vs))
1575                  (or (null excluded-script)
1576                      (null (setq vs (get-char-attribute variant 'script)))
1577                      (not (memq excluded-script vs))))
1578             (unless (and no-ucs-unified (get-char-attribute variant '=ucs))
1579               (insert-char-data variant printable)
1580               (if (setq ret (char-variants variant))
1581                   (while ret
1582                     (or (memq (car ret) variants)
1583                         ;; (get-char-attribute (car ret) '<-subsumptive)
1584                         (setq rest (nconc rest (list (car ret)))))
1585                     (setq ret (cdr ret)))))))
1586       (setq rest (cdr rest)))))
1587
1588 (defun insert-char-range-data (min max &optional script excluded-script)
1589   (let ((code min)
1590         char)
1591     (while (<= code max)
1592       (setq char (decode-char '=ucs code))
1593       (if (encode-char char '=ucs 'defined-only)
1594           (insert-char-data-with-variant char nil 'no-ucs-unified
1595                                          script excluded-script))
1596       (setq code (1+ code)))))
1597
1598 (defun write-char-range-data-to-file (min max file
1599                                           &optional script excluded-script)
1600   (let ((coding-system-for-write char-db-file-coding-system))
1601     (with-temp-buffer
1602       (insert (format ";; -*- coding: %s -*-\n"
1603                       char-db-file-coding-system))
1604       (insert-char-range-data min max script excluded-script)
1605       (write-region (point-min)(point-max) file))))
1606
1607 (defvar what-character-original-window-configuration)
1608
1609 ;;;###autoload
1610 (defun what-char-definition (char)
1611   (interactive (list (char-after)))
1612   (let ((buf (get-buffer-create "*Character Description*"))
1613         (the-buf (current-buffer))
1614         (win-conf (current-window-configuration)))
1615     (pop-to-buffer buf)
1616     (make-local-variable 'what-character-original-window-configuration)
1617     (setq what-character-original-window-configuration win-conf)
1618     (setq buffer-read-only nil)
1619     (erase-buffer)
1620     (condition-case err
1621         (progn
1622           (insert-char-data-with-variant char 'printable)
1623           (unless (char-attribute-alist char)
1624             (insert (format ";; = %c\n"
1625                             (let* ((rest (split-char char))
1626                                    (ccs (pop rest))
1627                                    (code (pop rest)))
1628                               (while rest
1629                                 (setq code (logior (lsh code 8)
1630                                                    (pop rest))))
1631                               (decode-char ccs code)))))
1632           ;; (char-db-update-comment)
1633           (set-buffer-modified-p nil)
1634           (view-mode the-buf (lambda (buf)
1635                                (set-window-configuration
1636                                 what-character-original-window-configuration)
1637                                ))
1638           (goto-char (point-min)))
1639       (error (progn
1640                (set-window-configuration
1641                 what-character-original-window-configuration)
1642                (signal (car err) (cdr err)))))))
1643
1644
1645 ;;; @ end
1646 ;;;
1647
1648 (provide 'char-db-util)
1649
1650 ;;; char-db-util.el ends here