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

   1 ;;; char-db-util.el --- Character Database utility -*- coding: utf-8-er; -*-
   2
   3 ;; Copyright (C) 1998, 1999, 2000, 2001, 2002, 2003, 2004, 2005, 2006, 2007,
   4 ;;   2008, 2009, 2010, 2011, 2012, 2013, 2014, 2015, 2016, 2017, 2018, 2019,
   5 ;;   2020, 2021, 2022 MORIOKA Tomohiko.
   6
   7 ;; Author: MORIOKA Tomohiko <tomo@kanji.zinbun.kyoto-u.ac.jp>
   8 ;; Keywords: CHISE, Character Database, ISO/IEC 10646, UCS, Unicode, MULE.
   9
  10 ;; This file is part of XEmacs CHISE.
  11
  12 ;; XEmacs CHISE is free software; you can redistribute it and/or
  13 ;; modify it under the terms of the GNU General Public License as
  14 ;; published by the Free Software Foundation; either version 2, or (at
  15 ;; your option) any later version.
  16
  17 ;; XEmacs CHISE is distributed in the hope that it will be useful,
  18 ;; but WITHOUT ANY WARRANTY; without even the implied warranty of
  19 ;; MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the GNU
  20 ;; General Public License for more details.
  21
  22 ;; You should have received a copy of the GNU General Public License
  23 ;; along with XEmacs CHISE; see the file COPYING.  If not, write to
  24 ;; the Free Software Foundation, Inc., 59 Temple Place - Suite 330,
  25 ;; Boston, MA 02111-1307, USA.
  26
  27 ;;; Code:
  28
  29 (require 'chise-subr)
  30 (require 'ideograph-subr)
  31
  32 (defconst unidata-normative-category-alist
  33   '(("Lu" letter        uppercase)
  34     ("Ll" letter        lowercase)
  35     ("Lt" letter        titlecase)
  36     ("Mn" mark          non-spacing)
  37     ("Mc" mark          spacing-combining)
  38     ("Me" mark          enclosing)
  39     ("Nd" number        decimal-digit)
  40     ("Nl" number        letter)
  41     ("No" number        other)
  42     ("Zs" separator     space)
  43     ("Zl" separator     line)
  44     ("Zp" separator     paragraph)
  45     ("Cc" other         control)
  46     ("Cf" other         format)
  47     ("Cs" other         surrogate)
  48     ("Co" other         private-use)
  49     ("Cn" other         not-assigned)))
  50
  51 (defconst unidata-informative-category-alist
  52   '(("Lm" letter        modifier)
  53     ("Lo" letter        other)
  54     ("Pc" punctuation   connector)
  55     ("Pd" punctuation   dash)
  56     ("Ps" punctuation   open)
  57     ("Pe" punctuation   close)
  58     ("Pi" punctuation   initial-quote)
  59     ("Pf" punctuation   final-quote)
  60     ("Po" punctuation   other)
  61     ("Sm" symbol        math)
  62     ("Sc" symbol        currency)
  63     ("Sk" symbol        modifier)
  64     ("So" symbol        other)
  65     ))
  66
  67 (defconst shuowen-radicals
  68   [?一 ?上 ?示 ?三 ?王 ?玉 ?玨 ?气 ?士 ?丨 ; 010
  69    ?屮 ?艸 ?蓐 ?茻 ?小 ?八 ?釆 ?半 ?牛 ?犛 ; 020
  70    ?告 ?口 ?凵 ?吅 ?哭 ?走 ?止 ?癶 ?步 ?此 ; 030
  71    ?正 ?是 ?辵 ?彳 ?廴 ?㢟 ?行 ?齒 ?牙 ?足 ; 040
  72    ?疋 ?品 ?龠 ?冊 ?㗊 ?舌 ?干 ?𧮫 ?只 ?㕯 ; 050
  73    ?句 ?丩 ?古 ?十 ?卅 ?言 ?誩 ?音 ?䇂 ?丵 ; 060
  74    ?菐 ?𠬞 ?𠬜 ?共 ?異 ?舁 ?𦥑 ?䢅 ?爨 ?革 ; 070
  75    ?鬲 ?䰜 ?爪 ?𠃨 ?鬥 ?又 ?𠂇 ?史 ?支 ?𦘒 ; 080
  76    ?聿 ?畫 ?隶 ?臤 ?臣 ?殳 ?殺 ?𠘧 ?寸 ?皮 ; 090
  77    ?㼱 ?攴 ?敎 ?卜 ?用 ?爻 ?㸚 ?𥄎 ?目 ?䀠 ; 100
  78    ?眉 ?盾 ?自 ?白 ?鼻 ?皕 ?習 ?羽 ?隹 ?奞 ; 110
  79    ?雈 ?𦫳 ?𥄕 ?羊 ?羴 ?瞿 ?雔 ?雥 ?鳥 ?烏 ; 120
  80    ?𠦒 ?冓 ?幺 ?𢆶 ?叀 ?玄 ?予 ?放 ?𠬪 ?𣦼 ; 130
  81    ?歺 ?死 ?冎 ?骨 ?肉 ?筋 ?刀 ?刃 ?㓞 ?丯 ; 140
  82    ?耒 ?角 ?竹 ?箕 ?丌 ?左 ?工 ?㠭 ?巫 ?甘 ; 150
  83    ?曰 ?乃 ?丂 ?可 ?兮 ?号 ?亏 ?旨 ?喜 ?壴 ; 160
  84 ;  ?旨 ?曰 ?乃 ?丂 ?可 ?兮 ?号 ?亏 ?喜 ?壴 ; 160
  85    ?鼓 ?豈 ?豆 ?豊 ?豐 ?䖒 ?虍 ?虎 ?虤 ?皿 ; 170
  86    ?𠙴 ?去 ?血 ?丶 ?丹 ?青 ?井 ?皀 ?鬯 ?食 ; 180
  87    ?亼 ?會 ?倉 ?入 ?缶 ?矢 ?高 ?冂 ?𩫏 ?京 ; 190
  88    ?亯 ?𣆪 ?畗 ?㐭 ?嗇 ?來 ?麥 ?夊 ?舛 ?䑞 ; 200
  89    ?韋 ?弟 ?夂 ?久 ?桀 ?木 ?東 ?林 ?才 ?叒 ; 210
  90    ?之 ?帀 ?出 ?𣎵 ?生 ?乇 ?𠂹 ?𠌶 ?華 ?𥝌 ; 220
  91    ?稽 ?巢 ?桼 ?束 ?㯻 ?囗 ?員 ?貝 ?邑 ?𨛜 ; 230
  92    ?日 ?旦 ?倝 ?㫃 ?冥 ?晶 ?月 ?有 ?明 ?囧 ; 240
  93    ?夕 ?多 ?毌 ?𢎘 ?𣐺 ?卣 ?齊 ?朿 ?片 ?鼎 ; 250
  94    ?克 ?彔 ?禾 ?秝 ?黍 ?香 ?米 ?毇 ?臼 ?凶 ; 260
  95    ?𣎳 ?𣏟 ?麻 ?尗 ?耑 ?韭 ?瓜 ?瓠 ?宀 ?宮 ; 270
  96    ?呂 ?穴 ?㝱 ?疒 ?冖 ?𠔼 ?冃 ?㒳 ?网 ?襾 ; 280
  97    ?巾 ?巿 ?帛 ?白 ?㡀 ?黹 ?人 ?𠤎 ?匕 ?从 ; 290
  98    ?比 ?北 ?丘 ?㐺 ?𡈼 ?重 ?臥 ?身 ?㐆 ?衣 ; 300
  99    ?裘 ?老 ?毛 ?毳 ?尸 ?尺 ?尾 ?履 ?舟 ?方 ; 310
 100    ?儿 ?兄 ?兂 ?皃 ?𠑹 ?先 ?秃 ?見 ?覞 ?欠 ; 320
 101    ?㱃 ?㳄 ?旡 ?頁 ?𦣻 ?面 ?丏 ?首 ?𥄉 ?須 ; 330
 102    ?彡 ?彣 ?文 ?髟 ?后 ?司 ?卮 ?卩 ?印 ?色 ; 340
 103    ?𠨍 ?辟 ?勹 ?包 ?茍 ?鬼 ?甶 ?厶 ?嵬 ?山 ; 350
 104    ?屾 ?屵 ?广 ?厂 ?丸 ?危 ?石 ?長 ?勿 ?冄 ; 360
 105    ?而 ?豕 ?㣇 ?彑 ?豚 ?豸 ?𤉡 ?易 ?象 ?馬 ; 370
 106    ?𢊁 ?鹿 ?麤 ?㲋 ?兔 ?萈 ?犬 ?㹜 ?鼠 ?能 ; 380
 107    ?熊 ?火 ?炎 ?黑 ?囪 ?焱 ?炙 ?赤 ?大 ?亦 ; 390
 108    ?夨 ?夭 ?交 ?尣 ?壺 ?壹 ?㚔 ?奢 ?亢 ?夲 ; 400
 109    ?夰 ?亣 ?夫 ?立 ?竝 ?囟 ?思 ?心 ?惢 ?水 ; 410
 110    ?沝 ?瀕 ?𡿨 ?巜 ?川 ?泉 ?灥 ?永 ?𠂢 ?谷 ; 420
 111    ?仌 ?雨 ?雲 ?魚 ?𩺰 ?燕 ?龍 ?飛 ?非 ?卂 ; 430
 112    ?𠃉 ?不 ?至 ?西 ?鹵 ?鹽 ?戶 ?門 ?耳 ?𦣝 ; 440
 113    ?手 ?𠦬 ?女 ?毋 ?民 ?丿 ?𠂆 ?乁 ?氏 ?氐 ; 450
 114    ?戈 ?戉 ?我 ?亅 ?琴 ?𠃊 ?亡 ?匸 ?匚 ?曲 ; 460
 115    ?甾 ?瓦 ?弓 ?弜 ?弦 ?系 ?糸 ?素 ?絲 ?率 ; 470
 116    ?虫 ?䖵 ?蟲 ?風 ?它 ?龜 ?黽 ?卵 ?二 ?土 ; 480
 117    ?垚 ?堇 ?里 ?田 ?畕 ?黃 ?男 ?力 ?劦 ?金 ; 490
 118    ?幵 ?勺 ?几 ?且 ?斤 ?斗 ?矛 ?車 ?𠂤 ?𨸏 ; 500
 119    ?𨺅 ?厽 ?四 ?宁 ?叕 ?亞 ?五 ?六 ?七 ?九 ; 510
 120    ?禸 ?嘼 ?甲 ?乙 ?丙 ?丁 ?戊 ?己 ?巴 ?庚 ; 520
 121    ?辛 ?辡 ?壬 ?癸 ?子 ?了 ?孨 ?𠫓 ?丑 ?寅 ; 530
 122    ?卯 ?辰 ?巳 ?午 ?未 ?申 ?酉 ?酋 ?戌 ?亥 ; 540
 123    ])
 124
 125 (defun shuowen-radical (number)
 126   (aref shuowen-radicals (1- number)))
 127
 128 (defvar char-db-file-coding-system 'utf-8-mcs-er)
 129
 130 (defvar char-db-ignored-attributes '(ideographic-products))
 131
 132 (defvar char-db-coded-charset-priority-list
 133   '(ascii
 134     control-1
 135     latin-iso8859-1
 136     latin-iso8859-2
 137     latin-iso8859-3
 138     latin-iso8859-4
 139     latin-iso8859-9
 140     latin-jisx0201
 141     cyrillic-iso8859-5
 142     greek-iso8859-7
 143     thai-tis620
 144     ;; =mj
 145     ;; =adobe-japan1-0
 146     ;; =adobe-japan1-1
 147     ;; =adobe-japan1-2
 148     ;; =adobe-japan1-3
 149     ;; =adobe-japan1-4
 150     ;; =adobe-japan1-5
 151     ;; =adobe-japan1-6
 152     =jis-x0208
 153     =jis-x0208@1978
 154     =jis-x0208@1983
 155     japanese-jisx0212
 156     chinese-gb2312
 157     =jis-x0208@1990
 158     chinese-cns11643-1
 159     chinese-cns11643-2
 160     chinese-cns11643-3
 161     chinese-cns11643-4
 162     chinese-cns11643-5
 163     chinese-cns11643-6
 164     chinese-cns11643-7
 165     =jis-x0213-1
 166     =jis-x0213-1@2000
 167     =jis-x0213-1@2004
 168     =jis-x0213-2
 169     korean-ksc5601
 170     chinese-isoir165
 171     katakana-jisx0201
 172     hebrew-iso8859-8
 173     chinese-gb12345
 174     latin-viscii
 175     ethiopic-ucs
 176     =big5-cdp
 177     =hanyo-denshi/ja
 178     =hanyo-denshi/jb
 179     =hanyo-denshi/jc
 180     =hanyo-denshi/jd
 181     =hanyo-denshi/ft
 182     =hanyo-denshi/ia
 183     =hanyo-denshi/ib
 184     =hanyo-denshi/hg
 185     =hanyo-denshi/jt
 186     =hanyo-denshi/ks
 187     =hanyo-denshi/tk
 188     ;; ==mj
 189     ;; ==adobe-japan1-0
 190     ;; ==adobe-japan1-1
 191     ;; ==adobe-japan1-2
 192     ;; ==adobe-japan1-3
 193     ;; ==adobe-japan1-4
 194     ;; ==adobe-japan1-5
 195     ;; ==adobe-japan1-6
 196     ==jis-x0208
 197     ==jis-x0213-1
 198     ==jis-x0213-2
 199     ==jis-x0212
 200     ==hanyo-denshi/ja
 201     ==hanyo-denshi/jb
 202     ==hanyo-denshi/jc
 203     ==hanyo-denshi/ft
 204     ==hanyo-denshi/ia
 205     ==hanyo-denshi/ib
 206     ==hanyo-denshi/hg
 207     ==hanyo-denshi/jt
 208     ==hanyo-denshi/ks
 209     ==hanyo-denshi/tk
 210     =gt
 211     =gt-k
 212     =daikanwa
 213     =daijiten
 214     =daikanwa@rev2
 215     =daikanwa@rev1
 216     =daikanwa/+p
 217     =daikanwa/ho
 218     ==daikanwa
 219     ==daijiten
 220     =cbeta
 221     ideograph-hanziku-1
 222     ideograph-hanziku-2
 223     ideograph-hanziku-3
 224     ideograph-hanziku-4
 225     ideograph-hanziku-5
 226     ideograph-hanziku-6
 227     ideograph-hanziku-7
 228     ideograph-hanziku-8
 229     ideograph-hanziku-9
 230     ideograph-hanziku-10
 231     ideograph-hanziku-11
 232     ideograph-hanziku-12
 233     ;; =>>>adobe-japan1-0
 234     ;; =>>>adobe-japan1-1
 235     ;; =>>>adobe-japan1-2
 236     ;; =>>>adobe-japan1-3
 237     ;; =>>>adobe-japan1-4
 238     ;; =>>>adobe-japan1-5
 239     ;; =>>>adobe-japan1-6
 240     ;; =>>>jis-x0208
 241     ;; =>>>jis-x0213-1
 242     ;; =>>>jis-x0213-2
 243     ;; =>>>hanyo-denshi/ja
 244     ;; =>>>hanyo-denshi/jb
 245     ;; =>>>hanyo-denshi/jc
 246     ;; =>>>hanyo-denshi/ft
 247     ;; =>>>hanyo-denshi/ib
 248     ;; =>>>hanyo-denshi/hg
 249     ;; =>>>hanyo-denshi/jt
 250     ;; =>>>hanyo-denshi/ks
 251     ;; =>>>gt
 252     =>>adobe-japan1-0
 253     =>>adobe-japan1-1
 254     =>>adobe-japan1-2
 255     =>>adobe-japan1-3
 256     =>>adobe-japan1-4
 257     =>>adobe-japan1-5
 258     =>>adobe-japan1-6
 259     =>>jis-x0208
 260     =>>jis-x0213-1
 261     =>>jis-x0213-1@2000
 262     =>>jis-x0213-1@2004
 263     =>>jis-x0213-2
 264     =>>jis-x0208@1978
 265     =>>hanyo-denshi/ft
 266     =>>hanyo-denshi/jt
 267     =>>hanyo-denshi/ks
 268     =>>gt
 269     =>>daikanwa
 270     =>>cbeta
 271     =+>jis-x0208
 272     =+>jis-x0213-1
 273     =+>jis-x0213-2
 274     =+>adobe-japan1-0
 275     =+>adobe-japan1-1
 276     =+>adobe-japan1-2
 277     =+>adobe-japan1-3
 278     =+>adobe-japan1-4
 279     =+>adobe-japan1-5
 280     =+>adobe-japan1-6
 281     =+>jis-x0208@1978
 282     =>jis-x0208
 283     =>jis-x0208@1997
 284     =>jis-x0213-1
 285     =>jis-x0213-1@2000
 286     =>jis-x0213-1@2004
 287     =>jis-x0213-2
 288     ==>ucs@bucs
 289     ==>daijiten
 290     ;; =>ucs@hanyo-denshi
 291     =>ucs@iso
 292     =>ucs@unicode
 293     =>ucs@jis
 294     =>ucs@cns
 295     =>ucs@ks
 296     =+>ucs@iso
 297     =+>ucs@unicode
 298     =+>ucs@jis
 299     =+>ucs@jis/1990
 300     =+>ucs@cns
 301     =+>ucs@ks
 302     =>>ucs@iso
 303     =>>ucs@unicode
 304     =>>ucs@jis
 305     =>>ucs@cns
 306     =>>>ucs@iso
 307     =>>>ucs@unicode
 308     ==ucs@iso
 309     ==ucs@unicode
 310     ;; ==ucs@cns
 311     ==gb2312
 312     ==ks-x1001
 313     ==cns11643-1
 314     ==cns11643-2
 315     ==cns11643-3
 316     ==cns11643-4
 317     ==cns11643-5
 318     ==cns11643-6
 319     ==cns11643-7
 320     ==gt
 321     ==jis-x0208@1990
 322     ;; ==jis-x0208@1983
 323     ==jis-x0208@1978
 324     ==gt-k
 325     =ucs@iso
 326     =ucs@unicode
 327     =ucs@cns
 328     ==big5-cdp
 329     ==cbeta
 330     =>>big5-cdp
 331     =>>gt-k
 332     =+>gt
 333     =+>big5-cdp
 334     =>big5-cdp
 335     =>big5-cdp@iwds-1
 336     =>mj
 337     =>hanyo-denshi/tk
 338     =>iwds-1
 339     =big5
 340     =big5-eten
 341     =zinbun-oracle
 342     =>zinbun-oracle
 343     =ruimoku-v6
 344     =>>ruimoku-v6
 345     ==ruimoku-v6
 346     =>ruimoku-v6
 347     =jef-china3
 348     =shinjigen
 349     =>cns11643-5
 350     =>cns11643-7
 351     =>gt
 352     =>gt-k
 353     =>daikanwa
 354     =>daikanwa/ho
 355     =>cbeta
 356     =ucs-var-001
 357     =ucs-var-002
 358     =ucs-var-003
 359     =ucs-var-004
 360     =ucs-var-005
 361     =ucs-var-006
 362     =ucs-var-007
 363     =ucs-var-008
 364     =ucs-var-010
 365     =ucs-itaiji-001
 366     =ucs-itaiji-002
 367     =ucs-itaiji-003
 368     =ucs-itaiji-004
 369     =ucs-itaiji-005
 370     =ucs-itaiji-006
 371     =ucs-itaiji-007
 372     =ucs-itaiji-008
 373     =ucs-itaiji-009
 374     =ucs-itaiji-010
 375     =ucs-itaiji-011
 376     =ucs-itaiji-012
 377     =ucs-itaiji-014
 378     =ucs-itaiji-016
 379     =+>ucs-var-003
 380     =>ucs-itaiji-001
 381     =>ucs-itaiji-002
 382     =>ucs-itaiji-003
 383     =>ucs-itaiji-004
 384     =>ucs-itaiji-005
 385     =>ucs-itaiji-006
 386     =>ucs-itaiji-007
 387     =>ucs-itaiji-008
 388     =>ucs-itaiji-009
 389     =>ucs-itaiji-010
 390     =>ucs-itaiji-011
 391     =>ucs-itaiji-012
 392     =>ucs-itaiji-014
 393     =>ucs-itaiji-016
 394     =big5-cdp-var-001
 395     =big5-cdp-var-002
 396     =big5-cdp-var-003
 397     =big5-cdp-var-004
 398     =big5-cdp-var-005
 399     =big5-cdp-var-007
 400     =big5-cdp-var-008
 401     =big5-cdp-var-010
 402     =big5-cdp-itaiji-001
 403     =big5-cdp-itaiji-002
 404     =big5-cdp-itaiji-003
 405     =big5-cdp-itaiji-004
 406     =big5-cdp-itaiji-005
 407     =>big5-cdp-itaiji-001
 408     =>big5-cdp-itaiji-002
 409     =>big5-cdp-itaiji-003
 410     =>big5-cdp-itaiji-004
 411     =>big5-cdp-var-001
 412     =>ucs@iwds-1
 413     =>ucs@cognate
 414     =>ucs@component
 415     =>ucs@iwds-1/normalized
 416     =>ucs-itaiji-001@iwds-1
 417     =>ucs-itaiji-002@iwds-1
 418     =>ucs-itaiji-003@iwds-1
 419     =>ucs-itaiji-006@iwds-1
 420     =>big5-cdp@component
 421     =>big5-cdp@cognate
 422     ==ucs@gb
 423     =ucs@gb
 424     ==ucs-var-002
 425     ==ucs-itaiji-002
 426     =ucs@JP/hanazono
 427     =koseki
 428     =shuowen-jiguge
 429     ===shuowen-jiguge5
 430     ===shuowen-jiguge4))
 431
 432
 433 ;;; @ char-db formatters
 434 ;;;
 435
 436 (defun char-db-make-char-spec (char)
 437   (let (ret char-spec)
 438     (cond ((characterp char)
 439            (cond ((and (setq ret (encode-char char '=ucs 'defined-only))
 440                        (not (and (<= #xE000 ret)(<= ret #xF8FF))))
 441                   (setq char-spec (list (cons '=ucs ret)))
 442                   (cond ((setq ret (get-char-attribute char 'name))
 443                          (setq char-spec (cons (cons 'name ret) char-spec))
 444                          )
 445                         ((setq ret (get-char-attribute char 'name*))
 446                          (setq char-spec (cons (cons 'name* ret) char-spec))
 447                          ))
 448                   )
 449                  ((encode-char char '=mj 'defined-only)
 450                   (setq char-spec nil)
 451                   (dolist (ccs (charset-list))
 452                     (if (and (or (eq ccs '=mj)
 453                                  ;; (eq (charset-property ccs 'iso-ir) 177)
 454                                  (string-match "=ucs@" (symbol-name ccs))
 455                                  )
 456                              (setq ccs (charset-name ccs))
 457                              (null (assq ccs char-spec))
 458                              (setq ret (encode-char char ccs 'defined-only)))
 459                         (setq char-spec (cons (cons ccs ret) char-spec))))
 460                   )
 461                  ((encode-char char '==mj 'defined-only)
 462                   (setq char-spec nil)
 463                   (dolist (ccs (charset-list))
 464                     (if (and (or (eq ccs '==mj)
 465                                  ;; (eq (charset-property ccs 'iso-ir) 177)
 466                                  (string-match "=ucs@" (symbol-name ccs))
 467                                  )
 468                              (setq ccs (charset-name ccs))
 469                              (null (assq ccs char-spec))
 470                              (setq ret (encode-char char ccs 'defined-only)))
 471                         (setq char-spec (cons (cons ccs ret) char-spec))))
 472                   )
 473                  ((encode-char char '=adobe-japan1 'defined-only)
 474                   (setq char-spec nil)
 475                   (dolist (ccs (charset-list))
 476                     (if (and (or (memq ccs
 477                                        '(=adobe-japan1-0
 478                                          =adobe-japan1-1
 479                                          =adobe-japan1-2
 480                                          =adobe-japan1-3
 481                                          =adobe-japan1-4
 482                                          =adobe-japan1-5
 483                                          =adobe-japan1-6
 484                                          ))
 485                                  ;; (eq (charset-property ccs 'iso-ir) 177)
 486                                  (string-match "=ucs@" (symbol-name ccs))
 487                                  )
 488                              (setq ccs (charset-name ccs))
 489                              (null (assq ccs char-spec))
 490                              (setq ret (encode-char char ccs 'defined-only)))
 491                         (setq char-spec (cons (cons ccs ret) char-spec))))
 492                   )
 493                  ((encode-char char '==adobe-japan1 'defined-only)
 494                   (setq char-spec nil)
 495                   (dolist (ccs (charset-list))
 496                     (if (and (or (memq ccs
 497                                        '(==adobe-japan1-0
 498                                          ==adobe-japan1-1
 499                                          ==adobe-japan1-2
 500                                          ==adobe-japan1-3
 501                                          ==adobe-japan1-4
 502                                          ==adobe-japan1-5
 503                                          ==adobe-japan1-6
 504                                          ))
 505                                  ;; (eq (charset-property ccs 'iso-ir) 177)
 506                                  (string-match "=ucs@" (symbol-name ccs))
 507                                  )
 508                              (setq ccs (charset-name ccs))
 509                              (null (assq ccs char-spec))
 510                              (setq ret (encode-char char ccs 'defined-only)))
 511                         (setq char-spec (cons (cons ccs ret) char-spec))))
 512                   )
 513                  ((setq ret
 514                         (catch 'tag
 515                           (let ((rest char-db-coded-charset-priority-list)
 516                                 ccs)
 517                             (while rest
 518                               (setq ccs (charset-name
 519                                          (find-charset (car rest))))
 520                               (if (setq ret
 521                                         (encode-char char ccs
 522                                                      'defined-only))
 523                                   (throw 'tag (cons ccs ret)))
 524                               (setq rest (cdr rest))))))
 525                   (setq char-spec (list ret))
 526                   (dolist (ccs (delq (car ret) (charset-list)))
 527                     (if (and (or (charset-iso-final-char ccs)
 528                                  (memq ccs
 529                                        '(=daikanwa
 530                                          =daikanwa@rev2
 531                                          ;; =gt-k
 532                                          =jis-x0208@1997
 533                                          ))
 534                                  (and (eq (charset-property ccs 'iso-ir) 177)
 535                                       (not (eq (charset-name ccs) '=ucs@big5)))
 536                                  ;; (string-match "=ucs@" (symbol-name ccs))
 537                                  )
 538                              (setq ccs (charset-name ccs))
 539                              (null (assq ccs char-spec))
 540                              (setq ret (encode-char char ccs 'defined-only)))
 541                         (setq char-spec (cons (cons ccs ret) char-spec))))
 542                   (if (null char-spec)
 543                       (setq char-spec (split-char char)))
 544                   (cond ((setq ret (get-char-attribute char 'name))
 545                          (setq char-spec (cons (cons 'name ret) char-spec))
 546                          )
 547                         ((setq ret (get-char-attribute char 'name*))
 548                          (setq char-spec (cons (cons 'name* ret) char-spec))
 549                          ))
 550                   )
 551                  ((setq ret (get-char-attribute
 552                              char 'ideographic-combination))
 553                   (setq char-spec
 554                         (cons (cons 'ideographic-combination ret)
 555                               char-spec))
 556                   ))
 557            char-spec)
 558           ((consp char)
 559            char))))
 560
 561 (defun char-db-insert-char-spec (char &optional readable column
 562                                       required-features)
 563   (unless column
 564     (setq column (current-column)))
 565   (let (char-spec temp-char)
 566     (setq char-spec (char-db-make-char-spec char))
 567     (unless (or (characterp char) ; char
 568                 (condition-case nil
 569                     (setq char (find-char char-spec))
 570                   (error nil)))
 571       ;; define temporary character
 572       ;;   Current implementation is dirty.
 573       (setq temp-char (define-char (cons '(ideograph-daikanwa . 0)
 574                                          char-spec)))
 575       (remove-char-attribute temp-char 'ideograph-daikanwa)
 576       (setq char temp-char))
 577     (insert-char-attributes char
 578                             readable
 579                             (union (mapcar #'car char-spec)
 580                                    required-features)
 581                             nil 'for-sub-node)
 582     (when temp-char
 583       ;; undefine temporary character
 584       ;;   Current implementation is dirty.
 585       (setq char-spec (char-attribute-alist temp-char))
 586       (while char-spec
 587         (remove-char-attribute temp-char (car (car char-spec)))
 588         (setq char-spec (cdr char-spec))))))
 589
 590 (defun char-db-insert-alist (alist &optional readable column)
 591   (unless column
 592     (setq column (current-column)))
 593   (let ((line-breaking
 594          (concat "\n" (make-string (1+ column) ?\ )))
 595         name value
 596         ret al ; cal
 597         key
 598         lbs cell rest separator)
 599     (insert "(")
 600     (while alist
 601       (setq name (car (car alist))
 602             value (cdr (car alist)))
 603       (cond ((eq name 'char)
 604              (insert "(char . ")
 605              (if (setq ret (condition-case nil
 606                                (find-char value)
 607                              (error nil)))
 608                  (progn
 609                    (setq al nil
 610                          ;; cal nil
 611                          )
 612                    (while value
 613                      (setq key (car (car value)))
 614                      ;; (if (find-charset key)
 615                      ;;     (setq cal (cons key cal))
 616                      (setq al (cons key al))
 617                      ;; )
 618                      (setq value (cdr value)))
 619                    (insert-char-attributes ret
 620                                            readable
 621                                            (or al 'none) ; cal
 622                                            nil 'for-sub-node))
 623                (insert (prin1-to-string value)))
 624              (insert ")")
 625              (insert line-breaking))
 626             ((consp value)
 627              (insert (format "(%-18s " name))
 628              (setq lbs (concat "\n" (make-string (current-column) ?\ )))
 629              (while (consp value)
 630                (setq cell (car value))
 631                (if (and (consp cell)
 632                         (consp (car cell))
 633                         (setq ret (condition-case nil
 634                                       (find-char cell)
 635                                     (error nil)))
 636                         )
 637                    (progn
 638                      (setq rest cell
 639                            al nil
 640                            ;; cal nil
 641                            )
 642                      (while rest
 643                        (setq key (car (car rest)))
 644                        ;; (if (find-charset key)
 645                        ;;     (setq cal (cons key cal))
 646                        (setq al (cons key al))
 647                        ;; )
 648                        (setq rest (cdr rest)))
 649                      (if separator
 650                          (insert lbs))
 651                      (insert-char-attributes ret
 652                                              readable
 653                                              al ; cal
 654                                              nil 'for-sub-node)
 655                      (setq separator lbs))
 656                  (if separator
 657                      (insert separator))
 658                  (insert (prin1-to-string cell))
 659                  (setq separator " "))
 660                (setq value (cdr value)))
 661              (insert ")")
 662              (insert line-breaking))
 663             (t
 664              (insert (format "(%-18s . %S)%s"
 665                              name value
 666                              line-breaking))))
 667       (setq alist (cdr alist))))
 668   (insert ")"))
 669
 670 (defun char-db-insert-char-reference (plist &optional readable column)
 671   (unless column
 672     (setq column (current-column)))
 673   (let ((line-breaking
 674          (concat "\n" (make-string (1+ column) ?\ )))
 675         (separator "")
 676         name value)
 677     (insert "(")
 678     (while plist
 679       (setq name (pop plist))
 680       (setq value (pop plist))
 681       (cond ((eq name :char)
 682              (insert separator)
 683              (insert ":char\t")
 684              (cond ((numberp value)
 685                     (setq value (decode-char '=ucs value)))
 686                    ;; ((consp value)
 687                    ;;  (setq value (or (find-char value)
 688                    ;;                  value)))
 689                    )
 690              (char-db-insert-char-spec value readable)
 691              (insert line-breaking)
 692              (setq separator ""))
 693             ((eq name :radical)
 694              (insert (format "%s%s\t%d ; %c%s"
 695                              separator
 696                              name value
 697                              (ideographic-radical value)
 698                              line-breaking))
 699              (setq separator ""))
 700             (t
 701              (insert (format "%s%s\t%S" separator name value))
 702              (setq separator line-breaking)))
 703       ))
 704   (insert ")"))
 705
 706 (defun char-db-decode-isolated-char (ccs code-point)
 707   (let (ret)
 708     (setq ret
 709           (cond ((eq ccs 'arabic-iso8859-6)
 710                  (decode-char ccs code-point))
 711                 ;; ((eq ccs '=gt)
 712                 ;;  (decode-builtin-char '==gt code-point))
 713                 ((and (memq ccs '(=gt-pj-1
 714                                   =gt-pj-2
 715                                   =gt-pj-3
 716                                   =gt-pj-4
 717                                   =gt-pj-5
 718                                   =gt-pj-6
 719                                   =gt-pj-7
 720                                   =gt-pj-8
 721                                   =gt-pj-9
 722                                   =gt-pj-10
 723                                   =gt-pj-11))
 724                       (setq ret (decode-char ccs code-point))
 725                       (setq ret (encode-char ret '=gt 'defined-only)))
 726                  (decode-builtin-char '=gt ret))
 727                 (t
 728                  (decode-builtin-char ccs code-point))))
 729     (cond ((null ret)
 730            (or (decode-char ccs code-point)
 731                (define-char (list (cons ccs code-point)))))
 732           ((and (<= 0 (char-int ret))
 733                 (<= (char-int ret) #x1F))
 734            (decode-char '=ucs (+ #x2400 (char-int ret))))
 735           ((= (char-int ret) #x7F)
 736            ?\u2421)
 737           (t ret))))
 738
 739 (defvar char-db-convert-obsolete-format t)
 740
 741 (defun char-db-insert-ccs-feature (name value line-breaking)
 742   (cond
 743    ((integerp value)
 744     (insert
 745      (format
 746       (cond
 747        ((memq name '(=>iwds-1
 748                      ==shinjigen
 749                      =shinjigen
 750                      =shinjigen@1ed ==shinjigen@1ed
 751                      =shinjigen@rev ==shinjigen@rev
 752                      =shinjigen/+p@rev ==shinjigen/+p@rev
 753                      ===daikanwa/ho ==daikanwa/ho
 754                      =daikanwa/ho =>>daikanwa/ho =>daikanwa/ho))
 755         "(%-18s .  %04d)\t; %c")
 756        ((eq name '=shinjigen@1ed/24pr)
 757         "(%-18s . %04d)\t; %c")
 758        ((or
 759          (memq name
 760                '(===daikanwa
 761                  ==daikanwa =daikanwa =>>daikanwa =>daikanwa
 762                  =daikanwa@rev1 =daikanwa@rev2
 763                  =daikanwa/+p ==daikanwa/+p ===daikanwa/+p
 764                  =>>daikanwa/+p
 765                  =daikanwa/+2p =>>daikanwa/+2p
 766                  =gt ==gt ===gt
 767                  =>>gt =+>gt =>gt
 768                  =gt-k ==gt-k ===gt-k
 769                  =>>gt-k =>gt-k
 770                  =adobe-japan1-0 ==adobe-japan1-0 ===adobe-japan1-0
 771                  =adobe-japan1-1 ==adobe-japan1-1 ===adobe-japan1-1
 772                  =adobe-japan1-2 ==adobe-japan1-2 ===adobe-japan1-2
 773                  =adobe-japan1-3 ==adobe-japan1-3 ===adobe-japan1-3
 774                  =adobe-japan1-4 ==adobe-japan1-4 ===adobe-japan1-4
 775                  =adobe-japan1-5 ==adobe-japan1-5 ===adobe-japan1-5
 776                  =adobe-japan1-6 ==adobe-japan1-6 ===adobe-japan1-6
 777                  =>>adobe-japan1-0 =+>adobe-japan1-0
 778                  =>>adobe-japan1-1 =+>adobe-japan1-1
 779                  =>>adobe-japan1-2 =+>adobe-japan1-2
 780                  =>>adobe-japan1-3 =+>adobe-japan1-3
 781                  =>>adobe-japan1-4 =+>adobe-japan1-4
 782                  =>>adobe-japan1-5 =+>adobe-japan1-5
 783                  =>>adobe-japan1-6 =+>adobe-japan1-6
 784                  =>cbeta =cbeta =>>cbeta ==cbeta ===cbeta
 785                  =zinbun-oracle =>zinbun-oracle
 786                  =daijiten ==daijiten ===daijiten ==>daijiten
 787                  ===hng-jou ===hng-keg ===hng-dng ===hng-mam
 788                  ===hng-drt ===hng-kgk ===hng-myz ===hng-kda
 789                  ===hng-khi ===hng-khm ===hng-fhs ===hng-hok
 790                  ===hng-kyd ===hng-sok
 791                  ===hng-yhk ===hng-kak ===hng-kar ===hng-kae
 792                  ===hng-sys ===hng-tsu ===hng-tzj
 793                  ===hng-hos ===hng-kkh ===hng-nak ===hng-jhk
 794                  ===hng-hod ===hng-gok ===hng-ink ===hng-nto
 795                  ===hng-nkm ===hng-k24 ===hng-ini ===hng-nkk
 796                  ===hng-kcc ===hng-kcj ===hng-kbk ===hng-sik
 797                  ===hng-skk ===hng-kyu ===hng-ksk ===hng-wan
 798                  ===hng-okd ===hng-wad ===hng-kmi
 799                  ===hng-zkd ===hng-doh ===hng-jyu ===hng-tzs
 800                  ===hng-sai ===hng-kad ===hng-kss
 801                  ===hng-kyo ===hng-ykk ===hng-saa
 802                  ===hng-sab ===hng-wks ===hng-wke ===hng-smk
 803                  ===hng-sgs ===hng-sts ===hng-kkd
 804                  =shuowen-jiguge ===shuowen-jiguge4 ===shuowen-jiguge5))
 805          ;; (string-match "^=adobe-" (symbol-name name))
 806          )
 807         "(%-18s . %05d)\t; %c")
 808        ((memq name '(=hanyo-denshi/ks
 809                      ==hanyo-denshi/ks ===hanyo-denshi/ks
 810                      =>>hanyo-denshi/ks
 811                      =koseki ==koseki
 812                      =mj ==mj ===mj =>>mj =>mj
 813                      =zihai mojikyo))
 814         "(%-18s . %06d)\t; %c")
 815        ((memq name '(=hanyo-denshi/tk =>hanyo-denshi/tk ==hanyo-denshi/tk))
 816         "(%-18s . %08d)\t; %c")
 817        ((>= (charset-dimension name) 2)
 818         "(%-18s . #x%04X)\t; %c")
 819        (t
 820         "(%-18s . #x%02X)\t; %c"))
 821       name
 822       (if (= (charset-iso-graphic-plane name) 1)
 823           (logior value
 824                   (cond ((= (charset-dimension name) 1)
 825                          #x80)
 826                         ((= (charset-dimension name) 2)
 827                          #x8080)
 828                         ((= (charset-dimension name) 3)
 829                          #x808080)
 830                         (t 0)))
 831         value)
 832       (char-db-decode-isolated-char name value)))
 833     (if (and (= (charset-chars name) 94)
 834              (= (charset-dimension name) 2))
 835         (insert (format " [%02d-%02d]"
 836                         (- (lsh value -8) 32)
 837                         (- (logand value 255) 32))))
 838     )
 839    (t
 840     (insert (format "(%-18s . %s)" name value))
 841     ))
 842   (insert line-breaking))
 843
 844 (defun char-db-insert-relation-feature (char name value line-breaking
 845                                              ccss readable)
 846   (insert (format "(%-18s%s " name line-breaking))
 847   (let ((lbs (concat "\n" (make-string (current-column) ?\ )))
 848         separator cell sources required-features
 849         ret)
 850     (while (consp value)
 851       (setq cell (car value))
 852       (if (integerp cell)
 853           (setq cell (decode-char '=ucs cell)))
 854       (cond
 855        ((eq name '->subsumptive)
 856         (when (or (not (some (lambda (atr)
 857                                (get-char-attribute cell atr))
 858                              char-db-ignored-attributes))
 859                   (some (lambda (ccs)
 860                           (encode-char cell ccs 'defined-only))
 861                         ccss))
 862           (if separator
 863               (insert lbs))
 864           (let ((char-db-ignored-attributes
 865                  (cons '<-subsumptive
 866                        char-db-ignored-attributes)))
 867             (insert-char-attributes cell readable nil nil 'for-sub-node))
 868           (setq separator lbs))
 869         )
 870        ((characterp cell)
 871         (setq sources
 872               (get-char-attribute
 873                char (intern (format "%s*sources" name))))
 874         (setq required-features nil)
 875         (dolist (source sources)
 876           (cond
 877            ((memq source '(JP
 878                            JP/Jouyou
 879                            shinjigen shinjigen@1ed shinjigen@rev))
 880             (setq required-features
 881                   (union required-features
 882                          '(=jis-x0208
 883                            =jis-x0208@1990
 884                            =jis-x0213-1@2000
 885                            =jis-x0213-1@2004
 886                            =jis-x0213-2
 887                            =jis-x0212
 888                            =jis-x0208@1983
 889                            =jis-x0208@1978
 890                            =shinjigen
 891                            =shinjigen@1ed
 892                            =shinjigen@rev
 893                            =shinjigen/+p@rev))))
 894            ((eq source 'CN)
 895             (setq required-features
 896                   (union required-features
 897                          '(=gb2312
 898                            =gb12345
 899                            =iso-ir165)))))
 900           (cond
 901            ((find-charset (setq ret (intern (format "=%s" source))))
 902             (setq required-features
 903                   (cons ret required-features)))
 904            (t (setq required-features
 905                     (cons source required-features)))))
 906         (cond ((string-match "@JP" (symbol-name name))
 907                (setq required-features
 908                      (union required-features
 909                             '(=jis-x0208
 910                               =jis-x0208@1990
 911                               =jis-x0213-1-2000
 912                               =jis-x0213-2-2000
 913                               =jis-x0212
 914                               =jis-x0208@1983
 915                               =jis-x0208@1978))))
 916               ((string-match "@CN" (symbol-name name))
 917                (setq required-features
 918                      (union required-features
 919                             '(=gb2312
 920                               =gb12345
 921                               =iso-ir165)))))
 922         (if separator
 923             (insert lbs))
 924         (if readable
 925             (insert (format "%S" cell))
 926           (char-db-insert-char-spec cell readable
 927                                     nil
 928                                     required-features))
 929         (setq separator lbs))
 930        ((consp cell)
 931         (if separator
 932             (insert lbs))
 933         (if (consp (car cell))
 934             (char-db-insert-char-spec cell readable)
 935           (char-db-insert-char-reference cell readable))
 936         (setq separator lbs))
 937        (t
 938         (if separator
 939             (insert separator))
 940         (insert (prin1-to-string cell))
 941         (setq separator " ")))
 942       (setq value (cdr value)))
 943     (insert ")")
 944     (insert line-breaking)))
 945
 946 (defun insert-char-attributes (char &optional readable attributes column
 947                                     for-sub-node)
 948   (unless column
 949     (setq column (current-column)))
 950   (let (name value ; has-long-ccs-name
 951         rest
 952         radical strokes
 953         (line-breaking
 954          (concat "\n" (make-string (1+ column) ?\ )))
 955         lbs cell separator ret
 956         key al cal
 957         dest-ccss ; sources required-features
 958         ccss)
 959     (let (atr-d)
 960       (setq attributes
 961             (sort (if attributes
 962                       (if (consp attributes)
 963                           (progn
 964                             (dolist (name attributes)
 965                               (unless (memq name char-db-ignored-attributes)
 966                                 (if (find-charset name)
 967                                     (push name ccss))
 968                                 (push name atr-d)))
 969                             atr-d))
 970                     (dolist (name (char-attribute-list))
 971                       (unless (memq name char-db-ignored-attributes)
 972                         (if (find-charset name)
 973                             (push name ccss))
 974                         (push name atr-d)))
 975                     atr-d)
 976                   #'char-attribute-name<)))
 977     (insert "(")
 978     (when (memq '<-subsumptive attributes)
 979       (when (or readable (not for-sub-node))
 980         (when (setq value (get-char-attribute char '<-subsumptive))
 981           (char-db-insert-relation-feature char '<-subsumptive value
 982                                            line-breaking
 983                                            ccss readable)))
 984       (setq attributes (delq '<-subsumptive attributes)))
 985     (when (and (memq '<-denotational attributes)
 986                (setq value (get-char-attribute char '<-denotational)))
 987       (char-db-insert-relation-feature char '<-denotational value
 988                                        line-breaking
 989                                        ccss readable)
 990       (setq attributes (delq '<-denotational attributes)))
 991     (when (and (memq '<-denotational@component attributes)
 992                (setq value (get-char-attribute char '<-denotational@component)))
 993       (char-db-insert-relation-feature char '<-denotational@component value
 994                                        line-breaking
 995                                        ccss readable)
 996       (setq attributes (delq '<-denotational@component attributes)))
 997     (when (and (memq '<-denotational@usage attributes)
 998                (setq value (get-char-attribute char '<-denotational@usage)))
 999       (char-db-insert-relation-feature char '<-denotational@usage value
1000                                        line-breaking
1001                                        ccss readable)
1002       (setq attributes (delq '<-denotational@usage attributes)))
1003     (when (and (memq 'name attributes)
1004                (setq value (get-char-attribute char 'name)))
1005       (insert (format
1006                (if (> (+ (current-column) (length value)) 48)
1007                    "(name . %S)%s"
1008                  "(name               . %S)%s")
1009                value line-breaking))
1010       (setq attributes (delq 'name attributes))
1011       )
1012     (when (and (memq 'name* attributes)
1013                (setq value (get-char-attribute char 'name*)))
1014       (insert (format
1015                (if (> (+ (current-column) (length value)) 48)
1016                    "(name* . %S)%s"
1017                  "(name*              . %S)%s")
1018                value line-breaking))
1019       (setq attributes (delq 'name* attributes))
1020       )
1021     (when (and (memq 'script attributes)
1022                (setq value (get-char-attribute char 'script)))
1023       (insert (format "(script\t\t%s)%s"
1024                       (mapconcat (function prin1-to-string)
1025                                  value " ")
1026                       line-breaking))
1027       (setq attributes (delq 'script attributes))
1028       )
1029     (dolist (name '(=>ucs =>ucs*))
1030       (when (and (memq name attributes)
1031                  (setq value (get-char-attribute char name)))
1032         (insert (format "(%-18s . #x%04X)\t; %c%s"
1033                         name value (decode-char '=ucs value)
1034                         line-breaking))
1035         (setq attributes (delq name attributes))))
1036     (dolist (name '(=>ucs@gb =>ucs@big5))
1037       (when (and (memq name attributes)
1038                  (setq value (get-char-attribute char name)))
1039         (insert (format "(%-18s . #x%04X)\t; %c%s"
1040                         name value
1041                         (decode-char (intern
1042                                       (concat "="
1043                                               (substring
1044                                                (symbol-name name) 2)))
1045                                      value)
1046                         line-breaking))
1047         (setq attributes (delq name attributes))
1048         ))
1049     ;; (dolist (name '(=>daikanwa))
1050     ;;   (when (and (memq name attributes)
1051     ;;              (setq value (get-char-attribute char name)))
1052     ;;     (insert
1053     ;;      (if (integerp value)
1054     ;;          (format "(%-18s . %05d)\t; %c%s"
1055     ;;                  name value (decode-char '=daikanwa value)
1056     ;;                  line-breaking)
1057     ;;        (format "(%-18s %s)\t; %c%s"
1058     ;;                name
1059     ;;                (mapconcat (function prin1-to-string)
1060     ;;                           value " ")
1061     ;;                (char-representative-of-daikanwa char)
1062     ;;                line-breaking)))
1063     ;;     (setq attributes (delq name attributes))))
1064     (when (and (memq 'general-category attributes)
1065                (setq value (get-char-attribute char 'general-category)))
1066       (insert (format
1067                "(general-category\t%s) ; %s%s"
1068                (mapconcat (lambda (cell)
1069                             (format "%S" cell))
1070                           value " ")
1071                (cond ((rassoc value unidata-normative-category-alist)
1072                       "Normative Category")
1073                      ((rassoc value unidata-informative-category-alist)
1074                       "Informative Category")
1075                      (t
1076                       "Unknown Category"))
1077                line-breaking))
1078       (setq attributes (delq 'general-category attributes))
1079       )
1080     (when (and (memq 'bidi-category attributes)
1081                (setq value (get-char-attribute char 'bidi-category)))
1082       (insert (format "(bidi-category\t. %S)%s"
1083                       value
1084                       line-breaking))
1085       (setq attributes (delq 'bidi-category attributes))
1086       )
1087     (unless (or (not (memq 'mirrored attributes))
1088                 (eq (setq value (get-char-attribute char 'mirrored 'empty))
1089                     'empty))
1090       (insert (format "(mirrored\t\t. %S)%s"
1091                       value
1092                       line-breaking))
1093       (setq attributes (delq 'mirrored attributes))
1094       )
1095     (cond
1096      ((and (memq 'decimal-digit-value attributes)
1097            (setq value (get-char-attribute char 'decimal-digit-value)))
1098       (insert (format "(decimal-digit-value . %S)%s"
1099                       value
1100                       line-breaking))
1101       (setq attributes (delq 'decimal-digit-value attributes))
1102       (when (and (memq 'digit-value attributes)
1103                  (setq value (get-char-attribute char 'digit-value)))
1104         (insert (format "(digit-value\t . %S)%s"
1105                         value
1106                         line-breaking))
1107         (setq attributes (delq 'digit-value attributes))
1108         )
1109       (when (and (memq 'numeric-value attributes)
1110                  (setq value (get-char-attribute char 'numeric-value)))
1111         (insert (format "(numeric-value\t . %S)%s"
1112                         value
1113                         line-breaking))
1114         (setq attributes (delq 'numeric-value attributes))
1115         )
1116       )
1117      (t
1118       (when (and (memq 'digit-value attributes)
1119                  (setq value (get-char-attribute char 'digit-value)))
1120         (insert (format "(digit-value\t. %S)%s"
1121                         value
1122                         line-breaking))
1123         (setq attributes (delq 'digit-value attributes))
1124         )
1125       (when (and (memq 'numeric-value attributes)
1126                  (setq value (get-char-attribute char 'numeric-value)))
1127         (insert (format "(numeric-value\t. %S)%s"
1128                         value
1129                         line-breaking))
1130         (setq attributes (delq 'numeric-value attributes))
1131         )))
1132     (when (and (memq 'iso-10646-comment attributes)
1133                (setq value (get-char-attribute char 'iso-10646-comment)))
1134       (insert (format "(iso-10646-comment\t. %S)%s"
1135                       value
1136                       line-breaking))
1137       (setq attributes (delq 'iso-10646-comment attributes))
1138       )
1139     (when (and (memq 'morohashi-daikanwa attributes)
1140                (setq value (get-char-attribute char 'morohashi-daikanwa)))
1141       (insert (format "(morohashi-daikanwa\t%s)%s"
1142                       (mapconcat (function prin1-to-string) value " ")
1143                       line-breaking))
1144       (setq attributes (delq 'morohashi-daikanwa attributes))
1145       )
1146     (setq radical nil
1147           strokes nil)
1148     (when (and (memq 'ideographic-radical attributes)
1149                (setq value (get-char-attribute char 'ideographic-radical)))
1150       (setq radical value)
1151       (insert (format "(ideographic-radical . %S)\t; %c%s"
1152                       radical
1153                       (ideographic-radical radical)
1154                       line-breaking))
1155       (setq attributes (delq 'ideographic-radical attributes))
1156       )
1157     (when (and (memq 'shuowen-radical attributes)
1158                (setq value (get-char-attribute char 'shuowen-radical)))
1159       (insert (format "(shuowen-radical\t. %S)\t; %c%s"
1160                       value
1161                       (shuowen-radical value)
1162                       line-breaking))
1163       (setq attributes (delq 'shuowen-radical attributes))
1164       )
1165     (let (key)
1166       (dolist (domain
1167                (append
1168                 char-db-feature-domains
1169                 (let (dest domain)
1170                   (dolist (feature (char-attribute-list))
1171                     (setq feature (symbol-name feature))
1172                     (when (string-match
1173                            "\\(radical\\|strokes\\)@\\([^@*]+\\)\\(\\*\\|$\\)"
1174                            feature)
1175                       (setq domain (intern (match-string 2 feature)))
1176                      (unless (memq domain dest)
1177                        (setq dest (cons domain dest)))))
1178                   (sort dest #'string<))))
1179         (setq key (intern (format "%s@%s" 'ideographic-radical domain)))
1180         (when (and (memq key attributes)
1181                    (setq value (get-char-attribute char key)))
1182           (setq radical value)
1183           (insert (format "(%s . %S)\t; %c%s"
1184                           key
1185                           radical
1186                           (ideographic-radical radical)
1187                           line-breaking))
1188           (setq attributes (delq key attributes))
1189           )
1190         (setq key (intern (format "%s@%s" 'ideographic-strokes domain)))
1191         (when (and (memq key attributes)
1192                    (setq value (get-char-attribute char key)))
1193           (setq strokes value)
1194           (insert (format "(%s . %S)%s"
1195                           key
1196                           strokes
1197                           line-breaking))
1198           (setq attributes (delq key attributes))
1199           )
1200         (setq key (intern (format "%s@%s" 'total-strokes domain)))
1201         (when (and (memq key attributes)
1202                    (setq value (get-char-attribute char key)))
1203           (insert (format "(%s       . %S)%s"
1204                           key
1205                           value
1206                           line-breaking))
1207           (setq attributes (delq key attributes))
1208           )
1209         (dolist (feature '(ideographic-radical
1210                            ideographic-strokes
1211                            total-strokes))
1212           (setq key (intern (format "%s@%s*sources" feature domain)))
1213           (when (and (memq key attributes)
1214                      (setq value (get-char-attribute char key)))
1215             (insert (format "(%s%s" key line-breaking))
1216             (dolist (cell value)
1217               (insert (format " %s" cell)))
1218             (insert ")")
1219             (insert line-breaking)
1220             (setq attributes (delq key attributes))
1221             ))
1222         ))
1223     (when (and (memq 'ideographic-strokes attributes)
1224                (setq value (get-char-attribute char 'ideographic-strokes)))
1225       (setq strokes value)
1226       (insert (format "(ideographic-strokes . %S)%s"
1227                       strokes
1228                       line-breaking))
1229       (setq attributes (delq 'ideographic-strokes attributes))
1230       )
1231     (when (and (memq 'kangxi-radical attributes)
1232                (setq value (get-char-attribute char 'kangxi-radical)))
1233       (unless (eq value radical)
1234         (insert (format "(kangxi-radical\t . %S)\t; %c%s"
1235                         value
1236                         (ideographic-radical value)
1237                         line-breaking))
1238         (or radical
1239             (setq radical value)))
1240       (setq attributes (delq 'kangxi-radical attributes))
1241       )
1242     (when (and (memq 'kangxi-strokes attributes)
1243                (setq value (get-char-attribute char 'kangxi-strokes)))
1244       (unless (eq value strokes)
1245         (insert (format "(kangxi-strokes\t . %S)%s"
1246                         value
1247                         line-breaking))
1248         (or strokes
1249             (setq strokes value)))
1250       (setq attributes (delq 'kangxi-strokes attributes))
1251       )
1252     ;; (when (and (memq 'japanese-radical attributes)
1253     ;;            (setq value (get-char-attribute char 'japanese-radical)))
1254     ;;   (unless (eq value radical)
1255     ;;     (insert (format "(japanese-radical\t . %S)\t; %c%s"
1256     ;;                     value
1257     ;;                     (ideographic-radical value)
1258     ;;                     line-breaking))
1259     ;;     (or radical
1260     ;;         (setq radical value)))
1261     ;;   (setq attributes (delq 'japanese-radical attributes))
1262     ;;   )
1263     (when (and (memq 'japanese-strokes attributes)
1264                (setq value (get-char-attribute char 'japanese-strokes)))
1265       (unless (eq value strokes)
1266         (insert (format "(japanese-strokes\t . %S)%s"
1267                         value
1268                         line-breaking))
1269         (or strokes
1270             (setq strokes value)))
1271       (setq attributes (delq 'japanese-strokes attributes))
1272       )
1273     (when (and (memq 'cns-radical attributes)
1274                (setq value (get-char-attribute char 'cns-radical)))
1275       (insert (format "(cns-radical\t . %S)\t; %c%s"
1276                       value
1277                       (ideographic-radical value)
1278                       line-breaking))
1279       (setq attributes (delq 'cns-radical attributes))
1280       )
1281     (when (and (memq 'cns-strokes attributes)
1282                (setq value (get-char-attribute char 'cns-strokes)))
1283       (unless (eq value strokes)
1284         (insert (format "(cns-strokes\t . %S)%s"
1285                         value
1286                         line-breaking))
1287         (or strokes
1288             (setq strokes value)))
1289       (setq attributes (delq 'cns-strokes attributes))
1290       )
1291     ;; (when (and (memq 'ideographic- attributes)
1292     ;;            (setq value (get-char-attribute char 'ideographic-)))
1293     ;;   (insert "(ideographic-       ")
1294     ;;   (setq lbs (concat "\n" (make-string (current-column) ?\ ))
1295     ;;         separator nil)
1296     ;;   (while (consp value)
1297     ;;     (setq cell (car value))
1298     ;;     (if (integerp cell)
1299     ;;         (setq cell (decode-char '=ucs cell)))
1300     ;;     (cond ((characterp cell)
1301     ;;            (if separator
1302     ;;                (insert lbs))
1303     ;;            (if readable
1304     ;;                (insert (format "%S" cell))
1305     ;;              (char-db-insert-char-spec cell readable))
1306     ;;            (setq separator lbs))
1307     ;;           ((consp cell)
1308     ;;            (if separator
1309     ;;                (insert lbs))
1310     ;;            (if (consp (car cell))
1311     ;;                (char-db-insert-char-spec cell readable)
1312     ;;              (char-db-insert-char-reference cell readable))
1313     ;;            (setq separator lbs))
1314     ;;           (t
1315     ;;            (if separator
1316     ;;                (insert separator))
1317     ;;            (insert (prin1-to-string cell))
1318     ;;            (setq separator " ")))
1319     ;;     (setq value (cdr value)))
1320     ;;   (insert ")")
1321     ;;   (insert line-breaking)
1322     ;;   (setq attributes (delq 'ideographic- attributes)))
1323     (when (and (memq 'total-strokes attributes)
1324                (setq value (get-char-attribute char 'total-strokes)))
1325       (insert (format "(total-strokes       . %S)%s"
1326                       value
1327                       line-breaking))
1328       (setq attributes (delq 'total-strokes attributes))
1329       )
1330     ;; (when (and (memq '->ideograph attributes)
1331     ;;            (setq value (get-char-attribute char '->ideograph)))
1332     ;;   (insert (format "(->ideograph\t%s)%s"
1333     ;;                   (mapconcat (lambda (code)
1334     ;;                                (cond ((symbolp code)
1335     ;;                                       (symbol-name code))
1336     ;;                                      ((integerp code)
1337     ;;                                       (format "#x%04X" code))
1338     ;;                                      (t
1339     ;;                                       (format "%s %S"
1340     ;;                                               line-breaking code))))
1341     ;;                              value " ")
1342     ;;                   line-breaking))
1343     ;;   (setq attributes (delq '->ideograph attributes))
1344     ;;   )
1345     ;; (when (and (memq '->decomposition attributes)
1346     ;;            (setq value (get-char-attribute char '->decomposition)))
1347     ;;   (insert (format "(->decomposition\t%s)%s"
1348     ;;                   (mapconcat (lambda (code)
1349     ;;                                (cond ((symbolp code)
1350     ;;                                       (symbol-name code))
1351     ;;                                      ((characterp code)
1352     ;;                                       (if readable
1353     ;;                                           (format "%S" code)
1354     ;;                                         (format "#x%04X"
1355     ;;                                                 (char-int code))
1356     ;;                                         ))
1357     ;;                                      ((integerp code)
1358     ;;                                       (format "#x%04X" code))
1359     ;;                                      (t
1360     ;;                                       (format "%s%S" line-breaking code))))
1361     ;;                              value " ")
1362     ;;                   line-breaking))
1363     ;;   (setq attributes (delq '->decomposition attributes))
1364     ;;   )
1365     (if (equal (get-char-attribute char '->titlecase)
1366                (get-char-attribute char '->uppercase))
1367         (setq attributes (delq '->titlecase attributes)))
1368     (when (and (memq '->mojikyo attributes)
1369                (setq value (get-char-attribute char '->mojikyo)))
1370       (insert (format "(->mojikyo\t\t. %06d)\t; %c%s"
1371                       value (decode-char 'mojikyo value)
1372                       line-breaking))
1373       (setq attributes (delq '->mojikyo attributes))
1374       )
1375     (unless readable
1376       (dolist (ignored '(composition
1377                          ->denotational <-subsumptive ->ucs-unified
1378                          ;; ->ideographic-component-forms
1379                          ))
1380         (setq attributes (delq ignored attributes))))
1381     (while attributes
1382       (setq name (car attributes))
1383       (unless (eq (setq value (get-char-attribute char name 'value-is-empty))
1384                   'value-is-empty)
1385         (cond ((setq ret (find-charset name))
1386                (setq name (charset-name ret))
1387                (when (not (memq name dest-ccss))
1388                  (setq dest-ccss (cons name dest-ccss))
1389                  (char-db-insert-ccs-feature name value line-breaking))
1390                )
1391               ((string-match "^=>ucs@" (symbol-name name))
1392                (insert (format "(%-18s . #x%04X)\t; %c%s"
1393                                name value (decode-char '=ucs value)
1394                                line-breaking))
1395                )
1396               ((eq name 'jisx0208-1978/4X)
1397                (insert (format "(%-18s . #x%04X)%s"
1398                                name value
1399                                line-breaking))
1400                )
1401               ((and
1402                 (not readable)
1403                 (not (eq name '->subsumptive))
1404                 (not (eq name '->uppercase))
1405                 (not (eq name '->lowercase))
1406                 (not (eq name '->titlecase))
1407                 (not (eq name '->canonical))
1408                 (not (eq name '->Bopomofo))
1409                 (not (eq name '->mistakable))
1410                 (not (eq name '->ideographic-variants))
1411                 (null (get-char-attribute
1412                        char (intern (format "%s*sources" name))))
1413                 (not (string-match "\\*sources$" (symbol-name name)))
1414                 (null (get-char-attribute
1415                        char (intern (format "%s*note" name))))
1416                 (not (string-match "\\*note$" (symbol-name name)))
1417                 (or (eq name '<-identical)
1418                     (eq name '<-uppercase)
1419                     (eq name '<-lowercase)
1420                     (eq name '<-titlecase)
1421                     (eq name '<-canonical)
1422                     (eq name '<-ideographic-variants)
1423                     ;; (eq name '<-synonyms)
1424                     (string-match "^<-synonyms" (symbol-name name))
1425                     (eq name '<-mistakable)
1426                     (when (string-match "^->" (symbol-name name))
1427                       (cond
1428                        ((string-match "^->fullwidth" (symbol-name name))
1429                         (not (and (consp value)
1430                                   (characterp (car value))
1431                                   (encode-char
1432                                    (car value) '=ucs 'defined-only)))
1433                         )
1434                        (t)))
1435                     ))
1436                )
1437               ((or (eq name 'ideographic-structure)
1438                    (eq name 'ideographic-structure@apparent)
1439                    (eq name 'ideographic-structure@apparent/leftmost)
1440                    (eq name 'ideographic-combination)
1441                    ;; (eq name 'ideographic-)
1442                    (eq name '=decomposition)
1443                    (char-feature-base-name= '=decomposition name)
1444                    (char-feature-base-name= '=>decomposition name)
1445                    ;; (string-match "^=>*decomposition\\(@[^*]+\\)?$"
1446                    ;;               (symbol-name name))
1447                    (string-match "^\\(->\\|<-\\)[^*]*$" (symbol-name name))
1448                    (string-match "^\\(->\\|<-\\)[^*]*\\*sources$"
1449                                  (symbol-name name))
1450                    )
1451                (char-db-insert-relation-feature char name value
1452                                                 line-breaking
1453                                                 ccss readable))
1454               ;; ((memq name '(ideograph=
1455               ;;               original-ideograph-of
1456               ;;               ancient-ideograph-of
1457               ;;               vulgar-ideograph-of
1458               ;;               wrong-ideograph-of
1459               ;;               ;; simplified-ideograph-of
1460               ;;               ideographic-variants
1461               ;;               ;; ideographic-different-form-of
1462               ;;               ))
1463               ;;  (insert (format "(%-18s%s " name line-breaking))
1464               ;;  (setq lbs (concat "\n" (make-string (current-column) ?\ ))
1465               ;;        separator nil)
1466               ;;  (while (consp value)
1467               ;;    (setq cell (car value))
1468               ;;    (if (and (consp cell)
1469               ;;             (consp (car cell)))
1470               ;;        (progn
1471               ;;          (if separator
1472               ;;              (insert lbs))
1473               ;;          (char-db-insert-alist cell readable)
1474               ;;          (setq separator lbs))
1475               ;;      (if separator
1476               ;;          (insert separator))
1477               ;;      (insert (prin1-to-string cell))
1478               ;;      (setq separator " "))
1479               ;;    (setq value (cdr value)))
1480               ;;  (insert ")")
1481               ;;  (insert line-breaking))
1482               ((consp value)
1483                (insert (format "(%-18s " name))
1484                (setq lbs (concat "\n" (make-string (current-column) ?\ ))
1485                      separator nil)
1486                (while (consp value)
1487                  (setq cell (car value))
1488                  (if (and (consp cell)
1489                           (consp (car cell))
1490                           (setq ret (condition-case nil
1491                                         (find-char cell)
1492                                       (error nil))))
1493                      (progn
1494                        (setq rest cell
1495                              al nil
1496                              cal nil)
1497                        (while rest
1498                          (setq key (car (car rest)))
1499                          (if (find-charset key)
1500                              (setq cal (cons key cal))
1501                            (setq al (cons key al)))
1502                          (setq rest (cdr rest)))
1503                        (if separator
1504                            (insert lbs))
1505                        (insert-char-attributes ret
1506                                                readable
1507                                                al ; cal
1508                                                nil 'for-sub-node)
1509                        (setq separator lbs))
1510                    (setq ret (prin1-to-string cell))
1511                    (if separator
1512                        (if (< (+ (current-column)
1513                                  (length ret)
1514                                  (length separator))
1515                               76)
1516                            (insert separator)
1517                          (insert lbs)))
1518                    (insert ret)
1519                    (setq separator " "))
1520                  (setq value (cdr value)))
1521                (insert ")")
1522                (insert line-breaking))
1523               (t
1524                (insert (format "(%-18s" name))
1525                (setq ret (prin1-to-string value))
1526                (unless (< (+ (current-column)
1527                              (length ret)
1528                              3)
1529                           76)
1530                  (insert line-breaking))
1531                (insert " . " ret ")" line-breaking)
1532                ;; (insert (format "(%-18s . %S)%s"
1533                ;;                 name value
1534                ;;                 line-breaking))
1535                )
1536               ))
1537       (setq attributes (cdr attributes)))
1538     (insert ")")))
1539
1540 (defun insert-char-data (char &optional readable
1541                               attributes)
1542   (save-restriction
1543     (narrow-to-region (point)(point))
1544     (insert "(define-char
1545   '")
1546     (insert-char-attributes char readable attributes)
1547     (insert ")\n")
1548     (goto-char (point-min))
1549     (while (re-search-forward "[ \t]+$" nil t)
1550       (replace-match ""))
1551     ;; from tabify.
1552     (goto-char (point-min))
1553     (while (re-search-forward "[ \t][ \t][ \t]*" nil t)
1554       (let ((column (current-column))
1555             (indent-tabs-mode t))
1556         (delete-region (match-beginning 0) (point))
1557         (indent-to column)))
1558     (goto-char (point-max))
1559     ;; (tabify (point-min)(point-max))
1560     ))
1561
1562 (defun insert-char-data-with-variant (char &optional printable
1563                                            no-ucs-unified
1564                                            script excluded-script)
1565   (insert-char-data char printable)
1566   (let ((variants (char-variants char))
1567         rest
1568         variant vs ret)
1569     (setq variants (sort variants #'<))
1570     (setq rest variants)
1571     (setq variants (cons char variants))
1572     (while rest
1573       (setq variant (car rest))
1574       (unless (get-char-attribute variant '<-subsumptive)
1575         (if (and (or (null script)
1576                      (null (setq vs (get-char-attribute variant 'script)))
1577                      (memq script vs))
1578                  (or (null excluded-script)
1579                      (null (setq vs (get-char-attribute variant 'script)))
1580                      (not (memq excluded-script vs))))
1581             (unless (and no-ucs-unified (get-char-attribute variant '=ucs))
1582               (insert-char-data variant printable)
1583               (if (setq ret (char-variants variant))
1584                   (while ret
1585                     (or (memq (car ret) variants)
1586                         ;; (get-char-attribute (car ret) '<-subsumptive)
1587                         (setq rest (nconc rest (list (car ret)))))
1588                     (setq ret (cdr ret)))))))
1589       (setq rest (cdr rest)))))
1590
1591 (defun insert-char-range-data (min max &optional script excluded-script)
1592   (let ((code min)
1593         char)
1594     (while (<= code max)
1595       (setq char (decode-char '=ucs code))
1596       (if (encode-char char '=ucs 'defined-only)
1597           (insert-char-data-with-variant char nil 'no-ucs-unified
1598                                          script excluded-script))
1599       (setq code (1+ code)))))
1600
1601 (defun write-char-range-data-to-file (min max file
1602                                           &optional script excluded-script)
1603   (let ((coding-system-for-write char-db-file-coding-system))
1604     (with-temp-buffer
1605       (insert (format ";; -*- coding: %s -*-\n"
1606                       char-db-file-coding-system))
1607       (insert-char-range-data min max script excluded-script)
1608       (write-region (point-min)(point-max) file))))
1609
1610 (defvar what-character-original-window-configuration)
1611
1612 ;;;###autoload
1613 (defun what-char-definition (char)
1614   (interactive (list (char-after)))
1615   (let ((buf (get-buffer-create "*Character Description*"))
1616         (the-buf (current-buffer))
1617         (win-conf (current-window-configuration)))
1618     (pop-to-buffer buf)
1619     (make-local-variable 'what-character-original-window-configuration)
1620     (setq what-character-original-window-configuration win-conf)
1621     (setq buffer-read-only nil)
1622     (erase-buffer)
1623     (condition-case err
1624         (progn
1625           (insert-char-data-with-variant char 'printable)
1626           (unless (char-attribute-alist char)
1627             (insert (format ";; = %c\n"
1628                             (let* ((rest (split-char char))
1629                                    (ccs (pop rest))
1630                                    (code (pop rest)))
1631                               (while rest
1632                                 (setq code (logior (lsh code 8)
1633                                                    (pop rest))))
1634                               (decode-char ccs code)))))
1635           ;; (char-db-update-comment)
1636           (set-buffer-modified-p nil)
1637           (view-mode the-buf (lambda (buf)
1638                                (set-window-configuration
1639                                 what-character-original-window-configuration)
1640                                ))
1641           (goto-char (point-min)))
1642       (error (progn
1643                (set-window-configuration
1644                 what-character-original-window-configuration)
1645                (signal (car err) (cdr err)))))))
1646
1647
1648 ;;; @ end
1649 ;;;
1650
1651 (provide 'char-db-util)
1652
1653 ;;; char-db-util.el ends here