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-008
 400     =big5-cdp-var-010
 401     =big5-cdp-itaiji-001
 402     =big5-cdp-itaiji-002
 403     =big5-cdp-itaiji-003
 404     =big5-cdp-itaiji-004
 405     =big5-cdp-itaiji-005
 406     =>big5-cdp-itaiji-001
 407     =>big5-cdp-itaiji-002
 408     =>big5-cdp-itaiji-003
 409     =>big5-cdp-itaiji-004
 410     =>big5-cdp-var-001
 411     =>ucs@iwds-1
 412     =>ucs@cognate
 413     =>ucs@component
 414     =>ucs@iwds-1/normalized
 415     =>ucs-itaiji-001@iwds-1
 416     =>ucs-itaiji-002@iwds-1
 417     =>ucs-itaiji-003@iwds-1
 418     =>ucs-itaiji-006@iwds-1
 419     =>big5-cdp@component
 420     =>big5-cdp@cognate
 421     ==ucs@gb
 422     =ucs@gb
 423     ==ucs-var-002
 424     =ucs@JP/hanazono
 425     =koseki
 426     =shuowen-jiguge
 427     ===shuowen-jiguge5
 428     ===shuowen-jiguge4))
 429
 430
 431 ;;; @ char-db formatters
 432 ;;;
 433
 434 (defun char-db-make-char-spec (char)
 435   (let (ret char-spec)
 436     (cond ((characterp char)
 437            (cond ((and (setq ret (encode-char char '=ucs 'defined-only))
 438                        (not (and (<= #xE000 ret)(<= ret #xF8FF))))
 439                   (setq char-spec (list (cons '=ucs ret)))
 440                   (cond ((setq ret (get-char-attribute char 'name))
 441                          (setq char-spec (cons (cons 'name ret) char-spec))
 442                          )
 443                         ((setq ret (get-char-attribute char 'name*))
 444                          (setq char-spec (cons (cons 'name* ret) char-spec))
 445                          ))
 446                   )
 447                  ((encode-char char '=mj 'defined-only)
 448                   (setq char-spec nil)
 449                   (dolist (ccs (charset-list))
 450                     (if (and (or (eq ccs '=mj)
 451                                  ;; (eq (charset-property ccs 'iso-ir) 177)
 452                                  (string-match "=ucs@" (symbol-name ccs))
 453                                  )
 454                              (setq ccs (charset-name ccs))
 455                              (null (assq ccs char-spec))
 456                              (setq ret (encode-char char ccs 'defined-only)))
 457                         (setq char-spec (cons (cons ccs ret) char-spec))))
 458                   )
 459                  ((encode-char char '==mj 'defined-only)
 460                   (setq char-spec nil)
 461                   (dolist (ccs (charset-list))
 462                     (if (and (or (eq ccs '==mj)
 463                                  ;; (eq (charset-property ccs 'iso-ir) 177)
 464                                  (string-match "=ucs@" (symbol-name ccs))
 465                                  )
 466                              (setq ccs (charset-name ccs))
 467                              (null (assq ccs char-spec))
 468                              (setq ret (encode-char char ccs 'defined-only)))
 469                         (setq char-spec (cons (cons ccs ret) char-spec))))
 470                   )
 471                  ((encode-char char '=adobe-japan1 'defined-only)
 472                   (setq char-spec nil)
 473                   (dolist (ccs (charset-list))
 474                     (if (and (or (memq ccs
 475                                        '(=adobe-japan1-0
 476                                          =adobe-japan1-1
 477                                          =adobe-japan1-2
 478                                          =adobe-japan1-3
 479                                          =adobe-japan1-4
 480                                          =adobe-japan1-5
 481                                          =adobe-japan1-6
 482                                          ))
 483                                  ;; (eq (charset-property ccs 'iso-ir) 177)
 484                                  (string-match "=ucs@" (symbol-name ccs))
 485                                  )
 486                              (setq ccs (charset-name ccs))
 487                              (null (assq ccs char-spec))
 488                              (setq ret (encode-char char ccs 'defined-only)))
 489                         (setq char-spec (cons (cons ccs ret) char-spec))))
 490                   )
 491                  ((encode-char char '==adobe-japan1 'defined-only)
 492                   (setq char-spec nil)
 493                   (dolist (ccs (charset-list))
 494                     (if (and (or (memq ccs
 495                                        '(==adobe-japan1-0
 496                                          ==adobe-japan1-1
 497                                          ==adobe-japan1-2
 498                                          ==adobe-japan1-3
 499                                          ==adobe-japan1-4
 500                                          ==adobe-japan1-5
 501                                          ==adobe-japan1-6
 502                                          ))
 503                                  ;; (eq (charset-property ccs 'iso-ir) 177)
 504                                  (string-match "=ucs@" (symbol-name ccs))
 505                                  )
 506                              (setq ccs (charset-name ccs))
 507                              (null (assq ccs char-spec))
 508                              (setq ret (encode-char char ccs 'defined-only)))
 509                         (setq char-spec (cons (cons ccs ret) char-spec))))
 510                   )
 511                  ((setq ret
 512                         (catch 'tag
 513                           (let ((rest char-db-coded-charset-priority-list)
 514                                 ccs)
 515                             (while rest
 516                               (setq ccs (charset-name
 517                                          (find-charset (car rest))))
 518                               (if (setq ret
 519                                         (encode-char char ccs
 520                                                      'defined-only))
 521                                   (throw 'tag (cons ccs ret)))
 522                               (setq rest (cdr rest))))))
 523                   (setq char-spec (list ret))
 524                   (dolist (ccs (delq (car ret) (charset-list)))
 525                     (if (and (or (charset-iso-final-char ccs)
 526                                  (memq ccs
 527                                        '(=daikanwa
 528                                          =daikanwa@rev2
 529                                          ;; =gt-k
 530                                          =jis-x0208@1997
 531                                          ))
 532                                  (and (eq (charset-property ccs 'iso-ir) 177)
 533                                       (not (eq (charset-name ccs) '=ucs@big5)))
 534                                  ;; (string-match "=ucs@" (symbol-name ccs))
 535                                  )
 536                              (setq ccs (charset-name ccs))
 537                              (null (assq ccs char-spec))
 538                              (setq ret (encode-char char ccs 'defined-only)))
 539                         (setq char-spec (cons (cons ccs ret) char-spec))))
 540                   (if (null char-spec)
 541                       (setq char-spec (split-char char)))
 542                   (cond ((setq ret (get-char-attribute char 'name))
 543                          (setq char-spec (cons (cons 'name ret) char-spec))
 544                          )
 545                         ((setq ret (get-char-attribute char 'name*))
 546                          (setq char-spec (cons (cons 'name* ret) char-spec))
 547                          ))
 548                   )
 549                  ((setq ret (get-char-attribute
 550                              char 'ideographic-combination))
 551                   (setq char-spec
 552                         (cons (cons 'ideographic-combination ret)
 553                               char-spec))
 554                   ))
 555            char-spec)
 556           ((consp char)
 557            char))))
 558
 559 (defun char-db-insert-char-spec (char &optional readable column
 560                                       required-features)
 561   (unless column
 562     (setq column (current-column)))
 563   (let (char-spec temp-char)
 564     (setq char-spec (char-db-make-char-spec char))
 565     (unless (or (characterp char) ; char
 566                 (condition-case nil
 567                     (setq char (find-char char-spec))
 568                   (error nil)))
 569       ;; define temporary character
 570       ;;   Current implementation is dirty.
 571       (setq temp-char (define-char (cons '(ideograph-daikanwa . 0)
 572                                          char-spec)))
 573       (remove-char-attribute temp-char 'ideograph-daikanwa)
 574       (setq char temp-char))
 575     (insert-char-attributes char
 576                             readable
 577                             (union (mapcar #'car char-spec)
 578                                    required-features)
 579                             nil 'for-sub-node)
 580     (when temp-char
 581       ;; undefine temporary character
 582       ;;   Current implementation is dirty.
 583       (setq char-spec (char-attribute-alist temp-char))
 584       (while char-spec
 585         (remove-char-attribute temp-char (car (car char-spec)))
 586         (setq char-spec (cdr char-spec))))))
 587
 588 (defun char-db-insert-alist (alist &optional readable column)
 589   (unless column
 590     (setq column (current-column)))
 591   (let ((line-breaking
 592          (concat "\n" (make-string (1+ column) ?\ )))
 593         name value
 594         ret al ; cal
 595         key
 596         lbs cell rest separator)
 597     (insert "(")
 598     (while alist
 599       (setq name (car (car alist))
 600             value (cdr (car alist)))
 601       (cond ((eq name 'char)
 602              (insert "(char . ")
 603              (if (setq ret (condition-case nil
 604                                (find-char value)
 605                              (error nil)))
 606                  (progn
 607                    (setq al nil
 608                          ;; cal nil
 609                          )
 610                    (while value
 611                      (setq key (car (car value)))
 612                      ;; (if (find-charset key)
 613                      ;;     (setq cal (cons key cal))
 614                      (setq al (cons key al))
 615                      ;; )
 616                      (setq value (cdr value)))
 617                    (insert-char-attributes ret
 618                                            readable
 619                                            (or al 'none) ; cal
 620                                            nil 'for-sub-node))
 621                (insert (prin1-to-string value)))
 622              (insert ")")
 623              (insert line-breaking))
 624             ((consp value)
 625              (insert (format "(%-18s " name))
 626              (setq lbs (concat "\n" (make-string (current-column) ?\ )))
 627              (while (consp value)
 628                (setq cell (car value))
 629                (if (and (consp cell)
 630                         (consp (car cell))
 631                         (setq ret (condition-case nil
 632                                       (find-char cell)
 633                                     (error nil)))
 634                         )
 635                    (progn
 636                      (setq rest cell
 637                            al nil
 638                            ;; cal nil
 639                            )
 640                      (while rest
 641                        (setq key (car (car rest)))
 642                        ;; (if (find-charset key)
 643                        ;;     (setq cal (cons key cal))
 644                        (setq al (cons key al))
 645                        ;; )
 646                        (setq rest (cdr rest)))
 647                      (if separator
 648                          (insert lbs))
 649                      (insert-char-attributes ret
 650                                              readable
 651                                              al ; cal
 652                                              nil 'for-sub-node)
 653                      (setq separator lbs))
 654                  (if separator
 655                      (insert separator))
 656                  (insert (prin1-to-string cell))
 657                  (setq separator " "))
 658                (setq value (cdr value)))
 659              (insert ")")
 660              (insert line-breaking))
 661             (t
 662              (insert (format "(%-18s . %S)%s"
 663                              name value
 664                              line-breaking))))
 665       (setq alist (cdr alist))))
 666   (insert ")"))
 667
 668 (defun char-db-insert-char-reference (plist &optional readable column)
 669   (unless column
 670     (setq column (current-column)))
 671   (let ((line-breaking
 672          (concat "\n" (make-string (1+ column) ?\ )))
 673         (separator "")
 674         name value)
 675     (insert "(")
 676     (while plist
 677       (setq name (pop plist))
 678       (setq value (pop plist))
 679       (cond ((eq name :char)
 680              (insert separator)
 681              (insert ":char\t")
 682              (cond ((numberp value)
 683                     (setq value (decode-char '=ucs value)))
 684                    ;; ((consp value)
 685                    ;;  (setq value (or (find-char value)
 686                    ;;                  value)))
 687                    )
 688              (char-db-insert-char-spec value readable)
 689              (insert line-breaking)
 690              (setq separator ""))
 691             ((eq name :radical)
 692              (insert (format "%s%s\t%d ; %c%s"
 693                              separator
 694                              name value
 695                              (ideographic-radical value)
 696                              line-breaking))
 697              (setq separator ""))
 698             (t
 699              (insert (format "%s%s\t%S" separator name value))
 700              (setq separator line-breaking)))
 701       ))
 702   (insert ")"))
 703
 704 (defun char-db-decode-isolated-char (ccs code-point)
 705   (let (ret)
 706     (setq ret
 707           (cond ((eq ccs 'arabic-iso8859-6)
 708                  (decode-char ccs code-point))
 709                 ;; ((eq ccs '=gt)
 710                 ;;  (decode-builtin-char '==gt code-point))
 711                 ((and (memq ccs '(=gt-pj-1
 712                                   =gt-pj-2
 713                                   =gt-pj-3
 714                                   =gt-pj-4
 715                                   =gt-pj-5
 716                                   =gt-pj-6
 717                                   =gt-pj-7
 718                                   =gt-pj-8
 719                                   =gt-pj-9
 720                                   =gt-pj-10
 721                                   =gt-pj-11))
 722                       (setq ret (decode-char ccs code-point))
 723                       (setq ret (encode-char ret '=gt 'defined-only)))
 724                  (decode-builtin-char '=gt ret))
 725                 (t
 726                  (decode-builtin-char ccs code-point))))
 727     (cond ((null ret)
 728            (or (decode-char ccs code-point)
 729                (define-char (list (cons ccs code-point)))))
 730           ((and (<= 0 (char-int ret))
 731                 (<= (char-int ret) #x1F))
 732            (decode-char '=ucs (+ #x2400 (char-int ret))))
 733           ((= (char-int ret) #x7F)
 734            ?\u2421)
 735           (t ret))))
 736
 737 (defvar char-db-convert-obsolete-format t)
 738
 739 (defun char-db-insert-ccs-feature (name value line-breaking)
 740   (cond
 741    ((integerp value)
 742     (insert
 743      (format
 744       (cond
 745        ((memq name '(=>iwds-1
 746                      ==shinjigen
 747                      =shinjigen
 748                      =shinjigen@1ed ==shinjigen@1ed
 749                      =shinjigen@rev ==shinjigen@rev
 750                      =shinjigen/+p@rev ==shinjigen/+p@rev
 751                      ===daikanwa/ho ==daikanwa/ho
 752                      =daikanwa/ho =>>daikanwa/ho =>daikanwa/ho))
 753         "(%-18s .  %04d)\t; %c")
 754        ((eq name '=shinjigen@1ed/24pr)
 755         "(%-18s . %04d)\t; %c")
 756        ((or
 757          (memq name
 758                '(===daikanwa
 759                  ==daikanwa =daikanwa =>>daikanwa =>daikanwa
 760                  =daikanwa@rev1 =daikanwa@rev2
 761                  =daikanwa/+p ==daikanwa/+p ===daikanwa/+p
 762                  =>>daikanwa/+p
 763                  =daikanwa/+2p =>>daikanwa/+2p
 764                  =gt ==gt ===gt
 765                  =>>gt =+>gt =>gt
 766                  =gt-k ==gt-k ===gt-k
 767                  =>>gt-k =>gt-k
 768                  =adobe-japan1-0 ==adobe-japan1-0 ===adobe-japan1-0
 769                  =adobe-japan1-1 ==adobe-japan1-1 ===adobe-japan1-1
 770                  =adobe-japan1-2 ==adobe-japan1-2 ===adobe-japan1-2
 771                  =adobe-japan1-3 ==adobe-japan1-3 ===adobe-japan1-3
 772                  =adobe-japan1-4 ==adobe-japan1-4 ===adobe-japan1-4
 773                  =adobe-japan1-5 ==adobe-japan1-5 ===adobe-japan1-5
 774                  =adobe-japan1-6 ==adobe-japan1-6 ===adobe-japan1-6
 775                  =>>adobe-japan1-0 =+>adobe-japan1-0
 776                  =>>adobe-japan1-1 =+>adobe-japan1-1
 777                  =>>adobe-japan1-2 =+>adobe-japan1-2
 778                  =>>adobe-japan1-3 =+>adobe-japan1-3
 779                  =>>adobe-japan1-4 =+>adobe-japan1-4
 780                  =>>adobe-japan1-5 =+>adobe-japan1-5
 781                  =>>adobe-japan1-6 =+>adobe-japan1-6
 782                  =>cbeta =cbeta =>>cbeta ==cbeta ===cbeta
 783                  =zinbun-oracle =>zinbun-oracle
 784                  =daijiten ==daijiten ===daijiten ==>daijiten
 785                  ===hng-jou ===hng-keg ===hng-dng ===hng-mam
 786                  ===hng-drt ===hng-kgk ===hng-myz ===hng-kda
 787                  ===hng-khi ===hng-khm ===hng-fhs ===hng-hok
 788                  ===hng-kyd ===hng-sok
 789                  ===hng-yhk ===hng-kak ===hng-kar ===hng-kae
 790                  ===hng-sys ===hng-tsu ===hng-tzj
 791                  ===hng-hos ===hng-kkh ===hng-nak ===hng-jhk
 792                  ===hng-hod ===hng-gok ===hng-ink ===hng-nto
 793                  ===hng-nkm ===hng-k24 ===hng-ini ===hng-nkk
 794                  ===hng-kcc ===hng-kcj ===hng-kbk ===hng-sik
 795                  ===hng-skk ===hng-kyu ===hng-ksk ===hng-wan
 796                  ===hng-okd ===hng-wad ===hng-kmi
 797                  ===hng-zkd ===hng-doh ===hng-jyu ===hng-tzs
 798                  ===hng-sai ===hng-kad ===hng-kss
 799                  ===hng-kyo ===hng-ykk ===hng-saa
 800                  ===hng-sab ===hng-wks ===hng-wke ===hng-smk
 801                  ===hng-sgs ===hng-sts ===hng-kkd
 802                  =shuowen-jiguge ===shuowen-jiguge4 ===shuowen-jiguge5))
 803          ;; (string-match "^=adobe-" (symbol-name name))
 804          )
 805         "(%-18s . %05d)\t; %c")
 806        ((memq name '(=hanyo-denshi/ks
 807                      ==hanyo-denshi/ks ===hanyo-denshi/ks
 808                      =>>hanyo-denshi/ks
 809                      =koseki ==koseki
 810                      =mj ==mj ===mj =>>mj =>mj
 811                      =zihai mojikyo))
 812         "(%-18s . %06d)\t; %c")
 813        ((memq name '(=hanyo-denshi/tk =>hanyo-denshi/tk ==hanyo-denshi/tk))
 814         "(%-18s . %08d)\t; %c")
 815        ((>= (charset-dimension name) 2)
 816         "(%-18s . #x%04X)\t; %c")
 817        (t
 818         "(%-18s . #x%02X)\t; %c"))
 819       name
 820       (if (= (charset-iso-graphic-plane name) 1)
 821           (logior value
 822                   (cond ((= (charset-dimension name) 1)
 823                          #x80)
 824                         ((= (charset-dimension name) 2)
 825                          #x8080)
 826                         ((= (charset-dimension name) 3)
 827                          #x808080)
 828                         (t 0)))
 829         value)
 830       (char-db-decode-isolated-char name value)))
 831     (if (and (= (charset-chars name) 94)
 832              (= (charset-dimension name) 2))
 833         (insert (format " [%02d-%02d]"
 834                         (- (lsh value -8) 32)
 835                         (- (logand value 255) 32))))
 836     )
 837    (t
 838     (insert (format "(%-18s . %s)" name value))
 839     ))
 840   (insert line-breaking))
 841
 842 (defun char-db-insert-relation-feature (char name value line-breaking
 843                                              ccss readable)
 844   (insert (format "(%-18s%s " name line-breaking))
 845   (let ((lbs (concat "\n" (make-string (current-column) ?\ )))
 846         separator cell sources required-features
 847         ret)
 848     (while (consp value)
 849       (setq cell (car value))
 850       (if (integerp cell)
 851           (setq cell (decode-char '=ucs cell)))
 852       (cond
 853        ((eq name '->subsumptive)
 854         (when (or (not (some (lambda (atr)
 855                                (get-char-attribute cell atr))
 856                              char-db-ignored-attributes))
 857                   (some (lambda (ccs)
 858                           (encode-char cell ccs 'defined-only))
 859                         ccss))
 860           (if separator
 861               (insert lbs))
 862           (let ((char-db-ignored-attributes
 863                  (cons '<-subsumptive
 864                        char-db-ignored-attributes)))
 865             (insert-char-attributes cell readable nil nil 'for-sub-node))
 866           (setq separator lbs))
 867         )
 868        ((characterp cell)
 869         (setq sources
 870               (get-char-attribute
 871                char (intern (format "%s*sources" name))))
 872         (setq required-features nil)
 873         (dolist (source sources)
 874           (cond
 875            ((memq source '(JP
 876                            JP/Jouyou
 877                            shinjigen shinjigen@1ed shinjigen@rev))
 878             (setq required-features
 879                   (union required-features
 880                          '(=jis-x0208
 881                            =jis-x0208@1990
 882                            =jis-x0213-1@2000
 883                            =jis-x0213-1@2004
 884                            =jis-x0213-2
 885                            =jis-x0212
 886                            =jis-x0208@1983
 887                            =jis-x0208@1978
 888                            =shinjigen
 889                            =shinjigen@1ed
 890                            =shinjigen@rev
 891                            =shinjigen/+p@rev))))
 892            ((eq source 'CN)
 893             (setq required-features
 894                   (union required-features
 895                          '(=gb2312
 896                            =gb12345
 897                            =iso-ir165)))))
 898           (cond
 899            ((find-charset (setq ret (intern (format "=%s" source))))
 900             (setq required-features
 901                   (cons ret required-features)))
 902            (t (setq required-features
 903                     (cons source required-features)))))
 904         (cond ((string-match "@JP" (symbol-name name))
 905                (setq required-features
 906                      (union required-features
 907                             '(=jis-x0208
 908                               =jis-x0208@1990
 909                               =jis-x0213-1-2000
 910                               =jis-x0213-2-2000
 911                               =jis-x0212
 912                               =jis-x0208@1983
 913                               =jis-x0208@1978))))
 914               ((string-match "@CN" (symbol-name name))
 915                (setq required-features
 916                      (union required-features
 917                             '(=gb2312
 918                               =gb12345
 919                               =iso-ir165)))))
 920         (if separator
 921             (insert lbs))
 922         (if readable
 923             (insert (format "%S" cell))
 924           (char-db-insert-char-spec cell readable
 925                                     nil
 926                                     required-features))
 927         (setq separator lbs))
 928        ((consp cell)
 929         (if separator
 930             (insert lbs))
 931         (if (consp (car cell))
 932             (char-db-insert-char-spec cell readable)
 933           (char-db-insert-char-reference cell readable))
 934         (setq separator lbs))
 935        (t
 936         (if separator
 937             (insert separator))
 938         (insert (prin1-to-string cell))
 939         (setq separator " ")))
 940       (setq value (cdr value)))
 941     (insert ")")
 942     (insert line-breaking)))
 943
 944 (defun insert-char-attributes (char &optional readable attributes column
 945                                     for-sub-node)
 946   (unless column
 947     (setq column (current-column)))
 948   (let (name value ; has-long-ccs-name
 949         rest
 950         radical strokes
 951         (line-breaking
 952          (concat "\n" (make-string (1+ column) ?\ )))
 953         lbs cell separator ret
 954         key al cal
 955         dest-ccss ; sources required-features
 956         ccss)
 957     (let (atr-d)
 958       (setq attributes
 959             (sort (if attributes
 960                       (if (consp attributes)
 961                           (progn
 962                             (dolist (name attributes)
 963                               (unless (memq name char-db-ignored-attributes)
 964                                 (if (find-charset name)
 965                                     (push name ccss))
 966                                 (push name atr-d)))
 967                             atr-d))
 968                     (dolist (name (char-attribute-list))
 969                       (unless (memq name char-db-ignored-attributes)
 970                         (if (find-charset name)
 971                             (push name ccss))
 972                         (push name atr-d)))
 973                     atr-d)
 974                   #'char-attribute-name<)))
 975     (insert "(")
 976     (when (memq '<-subsumptive attributes)
 977       (when (or readable (not for-sub-node))
 978         (when (setq value (get-char-attribute char '<-subsumptive))
 979           (char-db-insert-relation-feature char '<-subsumptive value
 980                                            line-breaking
 981                                            ccss readable)))
 982       (setq attributes (delq '<-subsumptive attributes)))
 983     (when (and (memq '<-denotational attributes)
 984                (setq value (get-char-attribute char '<-denotational)))
 985       (char-db-insert-relation-feature char '<-denotational value
 986                                        line-breaking
 987                                        ccss readable)
 988       (setq attributes (delq '<-denotational attributes)))
 989     (when (and (memq '<-denotational@component attributes)
 990                (setq value (get-char-attribute char '<-denotational@component)))
 991       (char-db-insert-relation-feature char '<-denotational@component value
 992                                        line-breaking
 993                                        ccss readable)
 994       (setq attributes (delq '<-denotational@component attributes)))
 995     (when (and (memq '<-denotational@usage attributes)
 996                (setq value (get-char-attribute char '<-denotational@usage)))
 997       (char-db-insert-relation-feature char '<-denotational@usage value
 998                                        line-breaking
 999                                        ccss readable)
1000       (setq attributes (delq '<-denotational@usage attributes)))
1001     (when (and (memq 'name attributes)
1002                (setq value (get-char-attribute char 'name)))
1003       (insert (format
1004                (if (> (+ (current-column) (length value)) 48)
1005                    "(name . %S)%s"
1006                  "(name               . %S)%s")
1007                value line-breaking))
1008       (setq attributes (delq 'name attributes))
1009       )
1010     (when (and (memq 'name* attributes)
1011                (setq value (get-char-attribute char 'name*)))
1012       (insert (format
1013                (if (> (+ (current-column) (length value)) 48)
1014                    "(name* . %S)%s"
1015                  "(name*              . %S)%s")
1016                value line-breaking))
1017       (setq attributes (delq 'name* attributes))
1018       )
1019     (when (and (memq 'script attributes)
1020                (setq value (get-char-attribute char 'script)))
1021       (insert (format "(script\t\t%s)%s"
1022                       (mapconcat (function prin1-to-string)
1023                                  value " ")
1024                       line-breaking))
1025       (setq attributes (delq 'script attributes))
1026       )
1027     (dolist (name '(=>ucs =>ucs*))
1028       (when (and (memq name attributes)
1029                  (setq value (get-char-attribute char name)))
1030         (insert (format "(%-18s . #x%04X)\t; %c%s"
1031                         name value (decode-char '=ucs value)
1032                         line-breaking))
1033         (setq attributes (delq name attributes))))
1034     (dolist (name '(=>ucs@gb =>ucs@big5))
1035       (when (and (memq name attributes)
1036                  (setq value (get-char-attribute char name)))
1037         (insert (format "(%-18s . #x%04X)\t; %c%s"
1038                         name value
1039                         (decode-char (intern
1040                                       (concat "="
1041                                               (substring
1042                                                (symbol-name name) 2)))
1043                                      value)
1044                         line-breaking))
1045         (setq attributes (delq name attributes))
1046         ))
1047     ;; (dolist (name '(=>daikanwa))
1048     ;;   (when (and (memq name attributes)
1049     ;;              (setq value (get-char-attribute char name)))
1050     ;;     (insert
1051     ;;      (if (integerp value)
1052     ;;          (format "(%-18s . %05d)\t; %c%s"
1053     ;;                  name value (decode-char '=daikanwa value)
1054     ;;                  line-breaking)
1055     ;;        (format "(%-18s %s)\t; %c%s"
1056     ;;                name
1057     ;;                (mapconcat (function prin1-to-string)
1058     ;;                           value " ")
1059     ;;                (char-representative-of-daikanwa char)
1060     ;;                line-breaking)))
1061     ;;     (setq attributes (delq name attributes))))
1062     (when (and (memq 'general-category attributes)
1063                (setq value (get-char-attribute char 'general-category)))
1064       (insert (format
1065                "(general-category\t%s) ; %s%s"
1066                (mapconcat (lambda (cell)
1067                             (format "%S" cell))
1068                           value " ")
1069                (cond ((rassoc value unidata-normative-category-alist)
1070                       "Normative Category")
1071                      ((rassoc value unidata-informative-category-alist)
1072                       "Informative Category")
1073                      (t
1074                       "Unknown Category"))
1075                line-breaking))
1076       (setq attributes (delq 'general-category attributes))
1077       )
1078     (when (and (memq 'bidi-category attributes)
1079                (setq value (get-char-attribute char 'bidi-category)))
1080       (insert (format "(bidi-category\t. %S)%s"
1081                       value
1082                       line-breaking))
1083       (setq attributes (delq 'bidi-category attributes))
1084       )
1085     (unless (or (not (memq 'mirrored attributes))
1086                 (eq (setq value (get-char-attribute char 'mirrored 'empty))
1087                     'empty))
1088       (insert (format "(mirrored\t\t. %S)%s"
1089                       value
1090                       line-breaking))
1091       (setq attributes (delq 'mirrored attributes))
1092       )
1093     (cond
1094      ((and (memq 'decimal-digit-value attributes)
1095            (setq value (get-char-attribute char 'decimal-digit-value)))
1096       (insert (format "(decimal-digit-value . %S)%s"
1097                       value
1098                       line-breaking))
1099       (setq attributes (delq 'decimal-digit-value attributes))
1100       (when (and (memq 'digit-value attributes)
1101                  (setq value (get-char-attribute char 'digit-value)))
1102         (insert (format "(digit-value\t . %S)%s"
1103                         value
1104                         line-breaking))
1105         (setq attributes (delq 'digit-value attributes))
1106         )
1107       (when (and (memq 'numeric-value attributes)
1108                  (setq value (get-char-attribute char 'numeric-value)))
1109         (insert (format "(numeric-value\t . %S)%s"
1110                         value
1111                         line-breaking))
1112         (setq attributes (delq 'numeric-value attributes))
1113         )
1114       )
1115      (t
1116       (when (and (memq 'digit-value attributes)
1117                  (setq value (get-char-attribute char 'digit-value)))
1118         (insert (format "(digit-value\t. %S)%s"
1119                         value
1120                         line-breaking))
1121         (setq attributes (delq 'digit-value attributes))
1122         )
1123       (when (and (memq 'numeric-value attributes)
1124                  (setq value (get-char-attribute char 'numeric-value)))
1125         (insert (format "(numeric-value\t. %S)%s"
1126                         value
1127                         line-breaking))
1128         (setq attributes (delq 'numeric-value attributes))
1129         )))
1130     (when (and (memq 'iso-10646-comment attributes)
1131                (setq value (get-char-attribute char 'iso-10646-comment)))
1132       (insert (format "(iso-10646-comment\t. %S)%s"
1133                       value
1134                       line-breaking))
1135       (setq attributes (delq 'iso-10646-comment attributes))
1136       )
1137     (when (and (memq 'morohashi-daikanwa attributes)
1138                (setq value (get-char-attribute char 'morohashi-daikanwa)))
1139       (insert (format "(morohashi-daikanwa\t%s)%s"
1140                       (mapconcat (function prin1-to-string) value " ")
1141                       line-breaking))
1142       (setq attributes (delq 'morohashi-daikanwa attributes))
1143       )
1144     (setq radical nil
1145           strokes nil)
1146     (when (and (memq 'ideographic-radical attributes)
1147                (setq value (get-char-attribute char 'ideographic-radical)))
1148       (setq radical value)
1149       (insert (format "(ideographic-radical . %S)\t; %c%s"
1150                       radical
1151                       (ideographic-radical radical)
1152                       line-breaking))
1153       (setq attributes (delq 'ideographic-radical attributes))
1154       )
1155     (when (and (memq 'shuowen-radical attributes)
1156                (setq value (get-char-attribute char 'shuowen-radical)))
1157       (insert (format "(shuowen-radical\t. %S)\t; %c%s"
1158                       value
1159                       (shuowen-radical value)
1160                       line-breaking))
1161       (setq attributes (delq 'shuowen-radical attributes))
1162       )
1163     (let (key)
1164       (dolist (domain
1165                (append
1166                 char-db-feature-domains
1167                 (let (dest domain)
1168                   (dolist (feature (char-attribute-list))
1169                     (setq feature (symbol-name feature))
1170                     (when (string-match
1171                            "\\(radical\\|strokes\\)@\\([^@*]+\\)\\(\\*\\|$\\)"
1172                            feature)
1173                       (setq domain (intern (match-string 2 feature)))
1174                      (unless (memq domain dest)
1175                        (setq dest (cons domain dest)))))
1176                   (sort dest #'string<))))
1177         (setq key (intern (format "%s@%s" 'ideographic-radical domain)))
1178         (when (and (memq key attributes)
1179                    (setq value (get-char-attribute char key)))
1180           (setq radical value)
1181           (insert (format "(%s . %S)\t; %c%s"
1182                           key
1183                           radical
1184                           (ideographic-radical radical)
1185                           line-breaking))
1186           (setq attributes (delq key attributes))
1187           )
1188         (setq key (intern (format "%s@%s" 'ideographic-strokes domain)))
1189         (when (and (memq key attributes)
1190                    (setq value (get-char-attribute char key)))
1191           (setq strokes value)
1192           (insert (format "(%s . %S)%s"
1193                           key
1194                           strokes
1195                           line-breaking))
1196           (setq attributes (delq key attributes))
1197           )
1198         (setq key (intern (format "%s@%s" 'total-strokes domain)))
1199         (when (and (memq key attributes)
1200                    (setq value (get-char-attribute char key)))
1201           (insert (format "(%s       . %S)%s"
1202                           key
1203                           value
1204                           line-breaking))
1205           (setq attributes (delq key attributes))
1206           )
1207         (dolist (feature '(ideographic-radical
1208                            ideographic-strokes
1209                            total-strokes))
1210           (setq key (intern (format "%s@%s*sources" feature domain)))
1211           (when (and (memq key attributes)
1212                      (setq value (get-char-attribute char key)))
1213             (insert (format "(%s%s" key line-breaking))
1214             (dolist (cell value)
1215               (insert (format " %s" cell)))
1216             (insert ")")
1217             (insert line-breaking)
1218             (setq attributes (delq key attributes))
1219             ))
1220         ))
1221     (when (and (memq 'ideographic-strokes attributes)
1222                (setq value (get-char-attribute char 'ideographic-strokes)))
1223       (setq strokes value)
1224       (insert (format "(ideographic-strokes . %S)%s"
1225                       strokes
1226                       line-breaking))
1227       (setq attributes (delq 'ideographic-strokes attributes))
1228       )
1229     (when (and (memq 'kangxi-radical attributes)
1230                (setq value (get-char-attribute char 'kangxi-radical)))
1231       (unless (eq value radical)
1232         (insert (format "(kangxi-radical\t . %S)\t; %c%s"
1233                         value
1234                         (ideographic-radical value)
1235                         line-breaking))
1236         (or radical
1237             (setq radical value)))
1238       (setq attributes (delq 'kangxi-radical attributes))
1239       )
1240     (when (and (memq 'kangxi-strokes attributes)
1241                (setq value (get-char-attribute char 'kangxi-strokes)))
1242       (unless (eq value strokes)
1243         (insert (format "(kangxi-strokes\t . %S)%s"
1244                         value
1245                         line-breaking))
1246         (or strokes
1247             (setq strokes value)))
1248       (setq attributes (delq 'kangxi-strokes attributes))
1249       )
1250     ;; (when (and (memq 'japanese-radical attributes)
1251     ;;            (setq value (get-char-attribute char 'japanese-radical)))
1252     ;;   (unless (eq value radical)
1253     ;;     (insert (format "(japanese-radical\t . %S)\t; %c%s"
1254     ;;                     value
1255     ;;                     (ideographic-radical value)
1256     ;;                     line-breaking))
1257     ;;     (or radical
1258     ;;         (setq radical value)))
1259     ;;   (setq attributes (delq 'japanese-radical attributes))
1260     ;;   )
1261     (when (and (memq 'japanese-strokes attributes)
1262                (setq value (get-char-attribute char 'japanese-strokes)))
1263       (unless (eq value strokes)
1264         (insert (format "(japanese-strokes\t . %S)%s"
1265                         value
1266                         line-breaking))
1267         (or strokes
1268             (setq strokes value)))
1269       (setq attributes (delq 'japanese-strokes attributes))
1270       )
1271     (when (and (memq 'cns-radical attributes)
1272                (setq value (get-char-attribute char 'cns-radical)))
1273       (insert (format "(cns-radical\t . %S)\t; %c%s"
1274                       value
1275                       (ideographic-radical value)
1276                       line-breaking))
1277       (setq attributes (delq 'cns-radical attributes))
1278       )
1279     (when (and (memq 'cns-strokes attributes)
1280                (setq value (get-char-attribute char 'cns-strokes)))
1281       (unless (eq value strokes)
1282         (insert (format "(cns-strokes\t . %S)%s"
1283                         value
1284                         line-breaking))
1285         (or strokes
1286             (setq strokes value)))
1287       (setq attributes (delq 'cns-strokes attributes))
1288       )
1289     ;; (when (and (memq 'ideographic- attributes)
1290     ;;            (setq value (get-char-attribute char 'ideographic-)))
1291     ;;   (insert "(ideographic-       ")
1292     ;;   (setq lbs (concat "\n" (make-string (current-column) ?\ ))
1293     ;;         separator nil)
1294     ;;   (while (consp value)
1295     ;;     (setq cell (car value))
1296     ;;     (if (integerp cell)
1297     ;;         (setq cell (decode-char '=ucs cell)))
1298     ;;     (cond ((characterp cell)
1299     ;;            (if separator
1300     ;;                (insert lbs))
1301     ;;            (if readable
1302     ;;                (insert (format "%S" cell))
1303     ;;              (char-db-insert-char-spec cell readable))
1304     ;;            (setq separator lbs))
1305     ;;           ((consp cell)
1306     ;;            (if separator
1307     ;;                (insert lbs))
1308     ;;            (if (consp (car cell))
1309     ;;                (char-db-insert-char-spec cell readable)
1310     ;;              (char-db-insert-char-reference cell readable))
1311     ;;            (setq separator lbs))
1312     ;;           (t
1313     ;;            (if separator
1314     ;;                (insert separator))
1315     ;;            (insert (prin1-to-string cell))
1316     ;;            (setq separator " ")))
1317     ;;     (setq value (cdr value)))
1318     ;;   (insert ")")
1319     ;;   (insert line-breaking)
1320     ;;   (setq attributes (delq 'ideographic- attributes)))
1321     (when (and (memq 'total-strokes attributes)
1322                (setq value (get-char-attribute char 'total-strokes)))
1323       (insert (format "(total-strokes       . %S)%s"
1324                       value
1325                       line-breaking))
1326       (setq attributes (delq 'total-strokes attributes))
1327       )
1328     ;; (when (and (memq '->ideograph attributes)
1329     ;;            (setq value (get-char-attribute char '->ideograph)))
1330     ;;   (insert (format "(->ideograph\t%s)%s"
1331     ;;                   (mapconcat (lambda (code)
1332     ;;                                (cond ((symbolp code)
1333     ;;                                       (symbol-name code))
1334     ;;                                      ((integerp code)
1335     ;;                                       (format "#x%04X" code))
1336     ;;                                      (t
1337     ;;                                       (format "%s %S"
1338     ;;                                               line-breaking code))))
1339     ;;                              value " ")
1340     ;;                   line-breaking))
1341     ;;   (setq attributes (delq '->ideograph attributes))
1342     ;;   )
1343     ;; (when (and (memq '->decomposition attributes)
1344     ;;            (setq value (get-char-attribute char '->decomposition)))
1345     ;;   (insert (format "(->decomposition\t%s)%s"
1346     ;;                   (mapconcat (lambda (code)
1347     ;;                                (cond ((symbolp code)
1348     ;;                                       (symbol-name code))
1349     ;;                                      ((characterp code)
1350     ;;                                       (if readable
1351     ;;                                           (format "%S" code)
1352     ;;                                         (format "#x%04X"
1353     ;;                                                 (char-int code))
1354     ;;                                         ))
1355     ;;                                      ((integerp code)
1356     ;;                                       (format "#x%04X" code))
1357     ;;                                      (t
1358     ;;                                       (format "%s%S" line-breaking code))))
1359     ;;                              value " ")
1360     ;;                   line-breaking))
1361     ;;   (setq attributes (delq '->decomposition attributes))
1362     ;;   )
1363     (if (equal (get-char-attribute char '->titlecase)
1364                (get-char-attribute char '->uppercase))
1365         (setq attributes (delq '->titlecase attributes)))
1366     (when (and (memq '->mojikyo attributes)
1367                (setq value (get-char-attribute char '->mojikyo)))
1368       (insert (format "(->mojikyo\t\t. %06d)\t; %c%s"
1369                       value (decode-char 'mojikyo value)
1370                       line-breaking))
1371       (setq attributes (delq '->mojikyo attributes))
1372       )
1373     (unless readable
1374       (dolist (ignored '(composition
1375                          ->denotational <-subsumptive ->ucs-unified
1376                          ;; ->ideographic-component-forms
1377                          ))
1378         (setq attributes (delq ignored attributes))))
1379     (while attributes
1380       (setq name (car attributes))
1381       (unless (eq (setq value (get-char-attribute char name 'value-is-empty))
1382                   'value-is-empty)
1383         (cond ((setq ret (find-charset name))
1384                (setq name (charset-name ret))
1385                (when (not (memq name dest-ccss))
1386                  (setq dest-ccss (cons name dest-ccss))
1387                  (char-db-insert-ccs-feature name value line-breaking))
1388                )
1389               ((string-match "^=>ucs@" (symbol-name name))
1390                (insert (format "(%-18s . #x%04X)\t; %c%s"
1391                                name value (decode-char '=ucs value)
1392                                line-breaking))
1393                )
1394               ((eq name 'jisx0208-1978/4X)
1395                (insert (format "(%-18s . #x%04X)%s"
1396                                name value
1397                                line-breaking))
1398                )
1399               ((and
1400                 (not readable)
1401                 (not (eq name '->subsumptive))
1402                 (not (eq name '->uppercase))
1403                 (not (eq name '->lowercase))
1404                 (not (eq name '->titlecase))
1405                 (not (eq name '->canonical))
1406                 (not (eq name '->Bopomofo))
1407                 (not (eq name '->mistakable))
1408                 (not (eq name '->ideographic-variants))
1409                 (null (get-char-attribute
1410                        char (intern (format "%s*sources" name))))
1411                 (not (string-match "\\*sources$" (symbol-name name)))
1412                 (null (get-char-attribute
1413                        char (intern (format "%s*note" name))))
1414                 (not (string-match "\\*note$" (symbol-name name)))
1415                 (or (eq name '<-identical)
1416                     (eq name '<-uppercase)
1417                     (eq name '<-lowercase)
1418                     (eq name '<-titlecase)
1419                     (eq name '<-canonical)
1420                     (eq name '<-ideographic-variants)
1421                     ;; (eq name '<-synonyms)
1422                     (string-match "^<-synonyms" (symbol-name name))
1423                     (eq name '<-mistakable)
1424                     (when (string-match "^->" (symbol-name name))
1425                       (cond
1426                        ((string-match "^->fullwidth" (symbol-name name))
1427                         (not (and (consp value)
1428                                   (characterp (car value))
1429                                   (encode-char
1430                                    (car value) '=ucs 'defined-only)))
1431                         )
1432                        (t)))
1433                     ))
1434                )
1435               ((or (eq name 'ideographic-structure)
1436                    (eq name 'ideographic-structure@apparent)
1437                    (eq name 'ideographic-structure@apparent/leftmost)
1438                    (eq name 'ideographic-combination)
1439                    ;; (eq name 'ideographic-)
1440                    (eq name '=decomposition)
1441                    (char-feature-base-name= '=decomposition name)
1442                    (char-feature-base-name= '=>decomposition name)
1443                    ;; (string-match "^=>*decomposition\\(@[^*]+\\)?$"
1444                    ;;               (symbol-name name))
1445                    (string-match "^\\(->\\|<-\\)[^*]*$" (symbol-name name))
1446                    (string-match "^\\(->\\|<-\\)[^*]*\\*sources$"
1447                                  (symbol-name name))
1448                    )
1449                (char-db-insert-relation-feature char name value
1450                                                 line-breaking
1451                                                 ccss readable))
1452               ;; ((memq name '(ideograph=
1453               ;;               original-ideograph-of
1454               ;;               ancient-ideograph-of
1455               ;;               vulgar-ideograph-of
1456               ;;               wrong-ideograph-of
1457               ;;               ;; simplified-ideograph-of
1458               ;;               ideographic-variants
1459               ;;               ;; ideographic-different-form-of
1460               ;;               ))
1461               ;;  (insert (format "(%-18s%s " name line-breaking))
1462               ;;  (setq lbs (concat "\n" (make-string (current-column) ?\ ))
1463               ;;        separator nil)
1464               ;;  (while (consp value)
1465               ;;    (setq cell (car value))
1466               ;;    (if (and (consp cell)
1467               ;;             (consp (car cell)))
1468               ;;        (progn
1469               ;;          (if separator
1470               ;;              (insert lbs))
1471               ;;          (char-db-insert-alist cell readable)
1472               ;;          (setq separator lbs))
1473               ;;      (if separator
1474               ;;          (insert separator))
1475               ;;      (insert (prin1-to-string cell))
1476               ;;      (setq separator " "))
1477               ;;    (setq value (cdr value)))
1478               ;;  (insert ")")
1479               ;;  (insert line-breaking))
1480               ((consp value)
1481                (insert (format "(%-18s " name))
1482                (setq lbs (concat "\n" (make-string (current-column) ?\ ))
1483                      separator nil)
1484                (while (consp value)
1485                  (setq cell (car value))
1486                  (if (and (consp cell)
1487                           (consp (car cell))
1488                           (setq ret (condition-case nil
1489                                         (find-char cell)
1490                                       (error nil))))
1491                      (progn
1492                        (setq rest cell
1493                              al nil
1494                              cal nil)
1495                        (while rest
1496                          (setq key (car (car rest)))
1497                          (if (find-charset key)
1498                              (setq cal (cons key cal))
1499                            (setq al (cons key al)))
1500                          (setq rest (cdr rest)))
1501                        (if separator
1502                            (insert lbs))
1503                        (insert-char-attributes ret
1504                                                readable
1505                                                al ; cal
1506                                                nil 'for-sub-node)
1507                        (setq separator lbs))
1508                    (setq ret (prin1-to-string cell))
1509                    (if separator
1510                        (if (< (+ (current-column)
1511                                  (length ret)
1512                                  (length separator))
1513                               76)
1514                            (insert separator)
1515                          (insert lbs)))
1516                    (insert ret)
1517                    (setq separator " "))
1518                  (setq value (cdr value)))
1519                (insert ")")
1520                (insert line-breaking))
1521               (t
1522                (insert (format "(%-18s" name))
1523                (setq ret (prin1-to-string value))
1524                (unless (< (+ (current-column)
1525                              (length ret)
1526                              3)
1527                           76)
1528                  (insert line-breaking))
1529                (insert " . " ret ")" line-breaking)
1530                ;; (insert (format "(%-18s . %S)%s"
1531                ;;                 name value
1532                ;;                 line-breaking))
1533                )
1534               ))
1535       (setq attributes (cdr attributes)))
1536     (insert ")")))
1537
1538 (defun insert-char-data (char &optional readable
1539                               attributes)
1540   (save-restriction
1541     (narrow-to-region (point)(point))
1542     (insert "(define-char
1543   '")
1544     (insert-char-attributes char readable attributes)
1545     (insert ")\n")
1546     (goto-char (point-min))
1547     (while (re-search-forward "[ \t]+$" nil t)
1548       (replace-match ""))
1549     ;; from tabify.
1550     (goto-char (point-min))
1551     (while (re-search-forward "[ \t][ \t][ \t]*" nil t)
1552       (let ((column (current-column))
1553             (indent-tabs-mode t))
1554         (delete-region (match-beginning 0) (point))
1555         (indent-to column)))
1556     (goto-char (point-max))
1557     ;; (tabify (point-min)(point-max))
1558     ))
1559
1560 (defun insert-char-data-with-variant (char &optional printable
1561                                            no-ucs-unified
1562                                            script excluded-script)
1563   (insert-char-data char printable)
1564   (let ((variants (char-variants char))
1565         rest
1566         variant vs ret)
1567     (setq variants (sort variants #'<))
1568     (setq rest variants)
1569     (setq variants (cons char variants))
1570     (while rest
1571       (setq variant (car rest))
1572       (unless (get-char-attribute variant '<-subsumptive)
1573         (if (and (or (null script)
1574                      (null (setq vs (get-char-attribute variant 'script)))
1575                      (memq script vs))
1576                  (or (null excluded-script)
1577                      (null (setq vs (get-char-attribute variant 'script)))
1578                      (not (memq excluded-script vs))))
1579             (unless (and no-ucs-unified (get-char-attribute variant '=ucs))
1580               (insert-char-data variant printable)
1581               (if (setq ret (char-variants variant))
1582                   (while ret
1583                     (or (memq (car ret) variants)
1584                         ;; (get-char-attribute (car ret) '<-subsumptive)
1585                         (setq rest (nconc rest (list (car ret)))))
1586                     (setq ret (cdr ret)))))))
1587       (setq rest (cdr rest)))))
1588
1589 (defun insert-char-range-data (min max &optional script excluded-script)
1590   (let ((code min)
1591         char)
1592     (while (<= code max)
1593       (setq char (decode-char '=ucs code))
1594       (if (encode-char char '=ucs 'defined-only)
1595           (insert-char-data-with-variant char nil 'no-ucs-unified
1596                                          script excluded-script))
1597       (setq code (1+ code)))))
1598
1599 (defun write-char-range-data-to-file (min max file
1600                                           &optional script excluded-script)
1601   (let ((coding-system-for-write char-db-file-coding-system))
1602     (with-temp-buffer
1603       (insert (format ";; -*- coding: %s -*-\n"
1604                       char-db-file-coding-system))
1605       (insert-char-range-data min max script excluded-script)
1606       (write-region (point-min)(point-max) file))))
1607
1608 (defvar what-character-original-window-configuration)
1609
1610 ;;;###autoload
1611 (defun what-char-definition (char)
1612   (interactive (list (char-after)))
1613   (let ((buf (get-buffer-create "*Character Description*"))
1614         (the-buf (current-buffer))
1615         (win-conf (current-window-configuration)))
1616     (pop-to-buffer buf)
1617     (make-local-variable 'what-character-original-window-configuration)
1618     (setq what-character-original-window-configuration win-conf)
1619     (setq buffer-read-only nil)
1620     (erase-buffer)
1621     (condition-case err
1622         (progn
1623           (insert-char-data-with-variant char 'printable)
1624           (unless (char-attribute-alist char)
1625             (insert (format ";; = %c\n"
1626                             (let* ((rest (split-char char))
1627                                    (ccs (pop rest))
1628                                    (code (pop rest)))
1629                               (while rest
1630                                 (setq code (logior (lsh code 8)
1631                                                    (pop rest))))
1632                               (decode-char ccs code)))))
1633           ;; (char-db-update-comment)
1634           (set-buffer-modified-p nil)
1635           (view-mode the-buf (lambda (buf)
1636                                (set-window-configuration
1637                                 what-character-original-window-configuration)
1638                                ))
1639           (goto-char (point-min)))
1640       (error (progn
1641                (set-window-configuration
1642                 what-character-original-window-configuration)
1643                (signal (car err) (cdr err)))))))
1644
1645
1646 ;;; @ end
1647 ;;;
1648
1649 (provide 'char-db-util)
1650
1651 ;;; char-db-util.el ends here