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 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     =>iwds-1
 338     =big5
 339     =big5-eten
 340     =zinbun-oracle
 341     =>zinbun-oracle
 342     =ruimoku-v6
 343     =>>ruimoku-v6
 344     ==ruimoku-v6
 345     =>ruimoku-v6
 346     =jef-china3
 347     =shinjigen
 348     =>cns11643-5
 349     =>cns11643-7
 350     =>gt
 351     =>gt-k
 352     =>daikanwa
 353     =>daikanwa/ho
 354     =>cbeta
 355     =ucs-var-001
 356     =ucs-var-002
 357     =ucs-var-003
 358     =ucs-var-004
 359     =ucs-var-005
 360     =ucs-var-006
 361     =ucs-var-007
 362     =ucs-var-008
 363     =ucs-var-010
 364     =ucs-itaiji-001
 365     =ucs-itaiji-002
 366     =ucs-itaiji-003
 367     =ucs-itaiji-004
 368     =ucs-itaiji-005
 369     =ucs-itaiji-006
 370     =ucs-itaiji-007
 371     =ucs-itaiji-008
 372     =ucs-itaiji-009
 373     =ucs-itaiji-010
 374     =ucs-itaiji-011
 375     =ucs-itaiji-012
 376     =ucs-itaiji-014
 377     =ucs-itaiji-016
 378     =+>ucs-var-003
 379     =>ucs-itaiji-001
 380     =>ucs-itaiji-002
 381     =>ucs-itaiji-003
 382     =>ucs-itaiji-004
 383     =>ucs-itaiji-005
 384     =>ucs-itaiji-006
 385     =>ucs-itaiji-007
 386     =>ucs-itaiji-008
 387     =>ucs-itaiji-009
 388     =>ucs-itaiji-010
 389     =>ucs-itaiji-011
 390     =>ucs-itaiji-014
 391     =big5-cdp-var-001
 392     =big5-cdp-var-002
 393     =big5-cdp-var-003
 394     =big5-cdp-var-004
 395     =big5-cdp-var-005
 396     =big5-cdp-var-007
 397     =big5-cdp-var-010
 398     =big5-cdp-itaiji-001
 399     =big5-cdp-itaiji-002
 400     =big5-cdp-itaiji-003
 401     =big5-cdp-itaiji-004
 402     =big5-cdp-itaiji-005
 403     =>big5-cdp-itaiji-001
 404     =>big5-cdp-itaiji-002
 405     =>big5-cdp-itaiji-003
 406     =>big5-cdp-itaiji-004
 407     =>big5-cdp-var-001
 408     =>ucs@iwds-1
 409     =>ucs@cognate
 410     =>ucs@component
 411     =>ucs@iwds-1/normalized
 412     =>ucs-itaiji-001@iwds-1
 413     =>ucs-itaiji-002@iwds-1
 414     =>ucs-itaiji-003@iwds-1
 415     =>ucs-itaiji-006@iwds-1
 416     =>big5-cdp@component
 417     =>big5-cdp@cognate
 418     ==ucs@gb
 419     =ucs@gb
 420     ==ucs-var-002
 421     =ucs@JP/hanazono
 422     =koseki
 423     =shuowen-jiguge
 424     ===shuowen-jiguge5
 425     ===shuowen-jiguge4))
 426
 427
 428 ;;; @ char-db formatters
 429 ;;;
 430
 431 (defun char-db-make-char-spec (char)
 432   (let (ret char-spec)
 433     (cond ((characterp char)
 434            (cond ((and (setq ret (encode-char char '=ucs 'defined-only))
 435                        (not (and (<= #xE000 ret)(<= ret #xF8FF))))
 436                   (setq char-spec (list (cons '=ucs ret)))
 437                   (cond ((setq ret (get-char-attribute char 'name))
 438                          (setq char-spec (cons (cons 'name ret) char-spec))
 439                          )
 440                         ((setq ret (get-char-attribute char 'name*))
 441                          (setq char-spec (cons (cons 'name* ret) char-spec))
 442                          ))
 443                   )
 444                  ((encode-char char '=mj 'defined-only)
 445                   (setq char-spec nil)
 446                   (dolist (ccs (charset-list))
 447                     (if (and (or (eq ccs '=mj)
 448                                  ;; (eq (charset-property ccs 'iso-ir) 177)
 449                                  (string-match "=ucs@" (symbol-name ccs))
 450                                  )
 451                              (setq ccs (charset-name ccs))
 452                              (null (assq ccs char-spec))
 453                              (setq ret (encode-char char ccs 'defined-only)))
 454                         (setq char-spec (cons (cons ccs ret) char-spec))))
 455                   )
 456                  ((encode-char char '==mj 'defined-only)
 457                   (setq char-spec nil)
 458                   (dolist (ccs (charset-list))
 459                     (if (and (or (eq ccs '==mj)
 460                                  ;; (eq (charset-property ccs 'iso-ir) 177)
 461                                  (string-match "=ucs@" (symbol-name ccs))
 462                                  )
 463                              (setq ccs (charset-name ccs))
 464                              (null (assq ccs char-spec))
 465                              (setq ret (encode-char char ccs 'defined-only)))
 466                         (setq char-spec (cons (cons ccs ret) char-spec))))
 467                   )
 468                  ((encode-char char '=adobe-japan1 'defined-only)
 469                   (setq char-spec nil)
 470                   (dolist (ccs (charset-list))
 471                     (if (and (or (memq ccs
 472                                        '(=adobe-japan1-0
 473                                          =adobe-japan1-1
 474                                          =adobe-japan1-2
 475                                          =adobe-japan1-3
 476                                          =adobe-japan1-4
 477                                          =adobe-japan1-5
 478                                          =adobe-japan1-6
 479                                          ))
 480                                  ;; (eq (charset-property ccs 'iso-ir) 177)
 481                                  (string-match "=ucs@" (symbol-name ccs))
 482                                  )
 483                              (setq ccs (charset-name ccs))
 484                              (null (assq ccs char-spec))
 485                              (setq ret (encode-char char ccs 'defined-only)))
 486                         (setq char-spec (cons (cons ccs ret) char-spec))))
 487                   )
 488                  ((encode-char char '==adobe-japan1 'defined-only)
 489                   (setq char-spec nil)
 490                   (dolist (ccs (charset-list))
 491                     (if (and (or (memq ccs
 492                                        '(==adobe-japan1-0
 493                                          ==adobe-japan1-1
 494                                          ==adobe-japan1-2
 495                                          ==adobe-japan1-3
 496                                          ==adobe-japan1-4
 497                                          ==adobe-japan1-5
 498                                          ==adobe-japan1-6
 499                                          ))
 500                                  ;; (eq (charset-property ccs 'iso-ir) 177)
 501                                  (string-match "=ucs@" (symbol-name ccs))
 502                                  )
 503                              (setq ccs (charset-name ccs))
 504                              (null (assq ccs char-spec))
 505                              (setq ret (encode-char char ccs 'defined-only)))
 506                         (setq char-spec (cons (cons ccs ret) char-spec))))
 507                   )
 508                  ((setq ret
 509                         (catch 'tag
 510                           (let ((rest char-db-coded-charset-priority-list)
 511                                 ccs)
 512                             (while rest
 513                               (setq ccs (charset-name
 514                                          (find-charset (car rest))))
 515                               (if (setq ret
 516                                         (encode-char char ccs
 517                                                      'defined-only))
 518                                   (throw 'tag (cons ccs ret)))
 519                               (setq rest (cdr rest))))))
 520                   (setq char-spec (list ret))
 521                   (dolist (ccs (delq (car ret) (charset-list)))
 522                     (if (and (or (charset-iso-final-char ccs)
 523                                  (memq ccs
 524                                        '(=daikanwa
 525                                          =daikanwa@rev2
 526                                          ;; =gt-k
 527                                          =jis-x0208@1997
 528                                          ))
 529                                  (and (eq (charset-property ccs 'iso-ir) 177)
 530                                       (not (eq (charset-name ccs) '=ucs@big5)))
 531                                  ;; (string-match "=ucs@" (symbol-name ccs))
 532                                  )
 533                              (setq ccs (charset-name ccs))
 534                              (null (assq ccs char-spec))
 535                              (setq ret (encode-char char ccs 'defined-only)))
 536                         (setq char-spec (cons (cons ccs ret) char-spec))))
 537                   (if (null char-spec)
 538                       (setq char-spec (split-char char)))
 539                   (cond ((setq ret (get-char-attribute char 'name))
 540                          (setq char-spec (cons (cons 'name ret) char-spec))
 541                          )
 542                         ((setq ret (get-char-attribute char 'name*))
 543                          (setq char-spec (cons (cons 'name* ret) char-spec))
 544                          ))
 545                   )
 546                  ((setq ret (get-char-attribute
 547                              char 'ideographic-combination))
 548                   (setq char-spec
 549                         (cons (cons 'ideographic-combination ret)
 550                               char-spec))
 551                   ))
 552            char-spec)
 553           ((consp char)
 554            char))))
 555
 556 (defun char-db-insert-char-spec (char &optional readable column
 557                                       required-features)
 558   (unless column
 559     (setq column (current-column)))
 560   (let (char-spec temp-char)
 561     (setq char-spec (char-db-make-char-spec char))
 562     (unless (or (characterp char) ; char
 563                 (condition-case nil
 564                     (setq char (find-char char-spec))
 565                   (error nil)))
 566       ;; define temporary character
 567       ;;   Current implementation is dirty.
 568       (setq temp-char (define-char (cons '(ideograph-daikanwa . 0)
 569                                          char-spec)))
 570       (remove-char-attribute temp-char 'ideograph-daikanwa)
 571       (setq char temp-char))
 572     (insert-char-attributes char
 573                             readable
 574                             (union (mapcar #'car char-spec)
 575                                    required-features)
 576                             nil 'for-sub-node)
 577     (when temp-char
 578       ;; undefine temporary character
 579       ;;   Current implementation is dirty.
 580       (setq char-spec (char-attribute-alist temp-char))
 581       (while char-spec
 582         (remove-char-attribute temp-char (car (car char-spec)))
 583         (setq char-spec (cdr char-spec))))))
 584
 585 (defun char-db-insert-alist (alist &optional readable column)
 586   (unless column
 587     (setq column (current-column)))
 588   (let ((line-breaking
 589          (concat "\n" (make-string (1+ column) ?\ )))
 590         name value
 591         ret al ; cal
 592         key
 593         lbs cell rest separator)
 594     (insert "(")
 595     (while alist
 596       (setq name (car (car alist))
 597             value (cdr (car alist)))
 598       (cond ((eq name 'char)
 599              (insert "(char . ")
 600              (if (setq ret (condition-case nil
 601                                (find-char value)
 602                              (error nil)))
 603                  (progn
 604                    (setq al nil
 605                          ;; cal nil
 606                          )
 607                    (while value
 608                      (setq key (car (car value)))
 609                      ;; (if (find-charset key)
 610                      ;;     (setq cal (cons key cal))
 611                      (setq al (cons key al))
 612                      ;; )
 613                      (setq value (cdr value)))
 614                    (insert-char-attributes ret
 615                                            readable
 616                                            (or al 'none) ; cal
 617                                            nil 'for-sub-node))
 618                (insert (prin1-to-string value)))
 619              (insert ")")
 620              (insert line-breaking))
 621             ((consp value)
 622              (insert (format "(%-18s " name))
 623              (setq lbs (concat "\n" (make-string (current-column) ?\ )))
 624              (while (consp value)
 625                (setq cell (car value))
 626                (if (and (consp cell)
 627                         (consp (car cell))
 628                         (setq ret (condition-case nil
 629                                       (find-char cell)
 630                                     (error nil)))
 631                         )
 632                    (progn
 633                      (setq rest cell
 634                            al nil
 635                            ;; cal nil
 636                            )
 637                      (while rest
 638                        (setq key (car (car rest)))
 639                        ;; (if (find-charset key)
 640                        ;;     (setq cal (cons key cal))
 641                        (setq al (cons key al))
 642                        ;; )
 643                        (setq rest (cdr rest)))
 644                      (if separator
 645                          (insert lbs))
 646                      (insert-char-attributes ret
 647                                              readable
 648                                              al ; cal
 649                                              nil 'for-sub-node)
 650                      (setq separator lbs))
 651                  (if separator
 652                      (insert separator))
 653                  (insert (prin1-to-string cell))
 654                  (setq separator " "))
 655                (setq value (cdr value)))
 656              (insert ")")
 657              (insert line-breaking))
 658             (t
 659              (insert (format "(%-18s . %S)%s"
 660                              name value
 661                              line-breaking))))
 662       (setq alist (cdr alist))))
 663   (insert ")"))
 664
 665 (defun char-db-insert-char-reference (plist &optional readable column)
 666   (unless column
 667     (setq column (current-column)))
 668   (let ((line-breaking
 669          (concat "\n" (make-string (1+ column) ?\ )))
 670         (separator "")
 671         name value)
 672     (insert "(")
 673     (while plist
 674       (setq name (pop plist))
 675       (setq value (pop plist))
 676       (cond ((eq name :char)
 677              (insert separator)
 678              (insert ":char\t")
 679              (cond ((numberp value)
 680                     (setq value (decode-char '=ucs value)))
 681                    ;; ((consp value)
 682                    ;;  (setq value (or (find-char value)
 683                    ;;                  value)))
 684                    )
 685              (char-db-insert-char-spec value readable)
 686              (insert line-breaking)
 687              (setq separator ""))
 688             ((eq name :radical)
 689              (insert (format "%s%s\t%d ; %c%s"
 690                              separator
 691                              name value
 692                              (ideographic-radical value)
 693                              line-breaking))
 694              (setq separator ""))
 695             (t
 696              (insert (format "%s%s\t%S" separator name value))
 697              (setq separator line-breaking)))
 698       ))
 699   (insert ")"))
 700
 701 (defun char-db-decode-isolated-char (ccs code-point)
 702   (let (ret)
 703     (setq ret
 704           (cond ((eq ccs 'arabic-iso8859-6)
 705                  (decode-char ccs code-point))
 706                 ;; ((eq ccs '=gt)
 707                 ;;  (decode-builtin-char '==gt code-point))
 708                 ((and (memq ccs '(=gt-pj-1
 709                                   =gt-pj-2
 710                                   =gt-pj-3
 711                                   =gt-pj-4
 712                                   =gt-pj-5
 713                                   =gt-pj-6
 714                                   =gt-pj-7
 715                                   =gt-pj-8
 716                                   =gt-pj-9
 717                                   =gt-pj-10
 718                                   =gt-pj-11))
 719                       (setq ret (decode-char ccs code-point))
 720                       (setq ret (encode-char ret '=gt 'defined-only)))
 721                  (decode-builtin-char '=gt ret))
 722                 (t
 723                  (decode-builtin-char ccs code-point))))
 724     (cond ((null ret)
 725            (or (decode-char ccs code-point)
 726                (define-char (list (cons ccs code-point)))))
 727           ((and (<= 0 (char-int ret))
 728                 (<= (char-int ret) #x1F))
 729            (decode-char '=ucs (+ #x2400 (char-int ret))))
 730           ((= (char-int ret) #x7F)
 731            ?\u2421)
 732           (t ret))))
 733
 734 (defvar char-db-convert-obsolete-format t)
 735
 736 (defun char-db-insert-ccs-feature (name value line-breaking)
 737   (cond
 738    ((integerp value)
 739     (insert
 740      (format
 741       (cond
 742        ((memq name '(=>iwds-1
 743                      ==shinjigen
 744                      =shinjigen
 745                      =shinjigen@1ed ==shinjigen@1ed
 746                      =shinjigen@rev ==shinjigen@rev
 747                      =shinjigen/+p@rev ==shinjigen/+p@rev
 748                      ===daikanwa/ho ==daikanwa/ho
 749                      =daikanwa/ho =>>daikanwa/ho =>daikanwa/ho))
 750         "(%-18s .  %04d)\t; %c")
 751        ((eq name '=shinjigen@1ed/24pr)
 752         "(%-18s . %04d)\t; %c")
 753        ((or
 754          (memq name
 755                '(===daikanwa
 756                  ==daikanwa =daikanwa =>>daikanwa =>daikanwa
 757                  =daikanwa@rev1 =daikanwa@rev2
 758                  =daikanwa/+p ==daikanwa/+p ===daikanwa/+p
 759                  =>>daikanwa/+p
 760                  =daikanwa/+2p =>>daikanwa/+2p
 761                  =gt ==gt ===gt
 762                  =>>gt =+>gt =>gt
 763                  =gt-k ==gt-k ===gt-k
 764                  =>>gt-k =>gt-k
 765                  =adobe-japan1-0 ==adobe-japan1-0 ===adobe-japan1-0
 766                  =adobe-japan1-1 ==adobe-japan1-1 ===adobe-japan1-1
 767                  =adobe-japan1-2 ==adobe-japan1-2 ===adobe-japan1-2
 768                  =adobe-japan1-3 ==adobe-japan1-3 ===adobe-japan1-3
 769                  =adobe-japan1-4 ==adobe-japan1-4 ===adobe-japan1-4
 770                  =adobe-japan1-5 ==adobe-japan1-5 ===adobe-japan1-5
 771                  =adobe-japan1-6 ==adobe-japan1-6 ===adobe-japan1-6
 772                  =>>adobe-japan1-0 =+>adobe-japan1-0
 773                  =>>adobe-japan1-1 =+>adobe-japan1-1
 774                  =>>adobe-japan1-2 =+>adobe-japan1-2
 775                  =>>adobe-japan1-3 =+>adobe-japan1-3
 776                  =>>adobe-japan1-4 =+>adobe-japan1-4
 777                  =>>adobe-japan1-5 =+>adobe-japan1-5
 778                  =>>adobe-japan1-6 =+>adobe-japan1-6
 779                  =>cbeta =cbeta =>>cbeta ==cbeta ===cbeta
 780                  =zinbun-oracle =>zinbun-oracle
 781                  =daijiten ==daijiten ===daijiten ==>daijiten
 782                  ===hng-jou ===hng-keg ===hng-dng ===hng-mam
 783                  ===hng-drt ===hng-kgk ===hng-myz ===hng-kda
 784                  ===hng-khi ===hng-khm ===hng-fhs ===hng-hok
 785                  ===hng-kyd ===hng-sok
 786                  ===hng-yhk ===hng-kak ===hng-kar ===hng-kae
 787                  ===hng-sys ===hng-tsu ===hng-tzj
 788                  ===hng-hos ===hng-kkh ===hng-nak ===hng-jhk
 789                  ===hng-hod ===hng-gok ===hng-ink ===hng-nto
 790                  ===hng-nkm ===hng-k24 ===hng-ini ===hng-nkk
 791                  ===hng-kcc ===hng-kcj ===hng-kbk ===hng-sik
 792                  ===hng-skk ===hng-kyu ===hng-ksk ===hng-wan
 793                  ===hng-okd ===hng-wad ===hng-kmi
 794                  ===hng-zkd ===hng-doh ===hng-jyu ===hng-tzs
 795                  ===hng-sai ===hng-kad ===hng-kss
 796                  ===hng-kyo ===hng-ykk ===hng-saa
 797                  ===hng-sab ===hng-wks ===hng-wke ===hng-smk
 798                  ===hng-sgs ===hng-sts ===hng-kkd
 799                  =shuowen-jiguge ===shuowen-jiguge4 ===shuowen-jiguge5))
 800          ;; (string-match "^=adobe-" (symbol-name name))
 801          )
 802         "(%-18s . %05d)\t; %c")
 803        ((memq name '(=hanyo-denshi/ks
 804                      ==hanyo-denshi/ks ===hanyo-denshi/ks
 805                      =>>hanyo-denshi/ks
 806                      =koseki ==koseki
 807                      =mj ==mj ===mj =>>mj =>mj
 808                      =zihai mojikyo))
 809         "(%-18s . %06d)\t; %c")
 810        ((memq name '(=hanyo-denshi/tk ==hanyo-denshi/tk))
 811         "(%-18s . %08d)\t; %c")
 812        ((>= (charset-dimension name) 2)
 813         "(%-18s . #x%04X)\t; %c")
 814        (t
 815         "(%-18s . #x%02X)\t; %c"))
 816       name
 817       (if (= (charset-iso-graphic-plane name) 1)
 818           (logior value
 819                   (cond ((= (charset-dimension name) 1)
 820                          #x80)
 821                         ((= (charset-dimension name) 2)
 822                          #x8080)
 823                         ((= (charset-dimension name) 3)
 824                          #x808080)
 825                         (t 0)))
 826         value)
 827       (char-db-decode-isolated-char name value)))
 828     (if (and (= (charset-chars name) 94)
 829              (= (charset-dimension name) 2))
 830         (insert (format " [%02d-%02d]"
 831                         (- (lsh value -8) 32)
 832                         (- (logand value 255) 32))))
 833     )
 834    (t
 835     (insert (format "(%-18s . %s)" name value))
 836     ))
 837   (insert line-breaking))
 838
 839 (defun char-db-insert-relation-feature (char name value line-breaking
 840                                              ccss readable)
 841   (insert (format "(%-18s%s " name line-breaking))
 842   (let ((lbs (concat "\n" (make-string (current-column) ?\ )))
 843         separator cell sources required-features
 844         ret)
 845     (while (consp value)
 846       (setq cell (car value))
 847       (if (integerp cell)
 848           (setq cell (decode-char '=ucs cell)))
 849       (cond
 850        ((eq name '->subsumptive)
 851         (when (or (not (some (lambda (atr)
 852                                (get-char-attribute cell atr))
 853                              char-db-ignored-attributes))
 854                   (some (lambda (ccs)
 855                           (encode-char cell ccs 'defined-only))
 856                         ccss))
 857           (if separator
 858               (insert lbs))
 859           (let ((char-db-ignored-attributes
 860                  (cons '<-subsumptive
 861                        char-db-ignored-attributes)))
 862             (insert-char-attributes cell readable nil nil 'for-sub-node))
 863           (setq separator lbs))
 864         )
 865        ((characterp cell)
 866         (setq sources
 867               (get-char-attribute
 868                char (intern (format "%s*sources" name))))
 869         (setq required-features nil)
 870         (dolist (source sources)
 871           (cond
 872            ((memq source '(JP
 873                            JP/Jouyou
 874                            shinjigen shinjigen@1ed shinjigen@rev))
 875             (setq required-features
 876                   (union required-features
 877                          '(=jis-x0208
 878                            =jis-x0208@1990
 879                            =jis-x0213-1@2000
 880                            =jis-x0213-1@2004
 881                            =jis-x0213-2
 882                            =jis-x0212
 883                            =jis-x0208@1983
 884                            =jis-x0208@1978
 885                            =shinjigen
 886                            =shinjigen@1ed
 887                            =shinjigen@rev
 888                            =shinjigen/+p@rev))))
 889            ((eq source 'CN)
 890             (setq required-features
 891                   (union required-features
 892                          '(=gb2312
 893                            =gb12345
 894                            =iso-ir165)))))
 895           (cond
 896            ((find-charset (setq ret (intern (format "=%s" source))))
 897             (setq required-features
 898                   (cons ret required-features)))
 899            (t (setq required-features
 900                     (cons source required-features)))))
 901         (cond ((string-match "@JP" (symbol-name name))
 902                (setq required-features
 903                      (union required-features
 904                             '(=jis-x0208
 905                               =jis-x0208@1990
 906                               =jis-x0213-1-2000
 907                               =jis-x0213-2-2000
 908                               =jis-x0212
 909                               =jis-x0208@1983
 910                               =jis-x0208@1978))))
 911               ((string-match "@CN" (symbol-name name))
 912                (setq required-features
 913                      (union required-features
 914                             '(=gb2312
 915                               =gb12345
 916                               =iso-ir165)))))
 917         (if separator
 918             (insert lbs))
 919         (if readable
 920             (insert (format "%S" cell))
 921           (char-db-insert-char-spec cell readable
 922                                     nil
 923                                     required-features))
 924         (setq separator lbs))
 925        ((consp cell)
 926         (if separator
 927             (insert lbs))
 928         (if (consp (car cell))
 929             (char-db-insert-char-spec cell readable)
 930           (char-db-insert-char-reference cell readable))
 931         (setq separator lbs))
 932        (t
 933         (if separator
 934             (insert separator))
 935         (insert (prin1-to-string cell))
 936         (setq separator " ")))
 937       (setq value (cdr value)))
 938     (insert ")")
 939     (insert line-breaking)))
 940
 941 (defun insert-char-attributes (char &optional readable attributes column
 942                                     for-sub-node)
 943   (unless column
 944     (setq column (current-column)))
 945   (let (name value ; has-long-ccs-name
 946         rest
 947         radical strokes
 948         (line-breaking
 949          (concat "\n" (make-string (1+ column) ?\ )))
 950         lbs cell separator ret
 951         key al cal
 952         dest-ccss ; sources required-features
 953         ccss)
 954     (let (atr-d)
 955       (setq attributes
 956             (sort (if attributes
 957                       (if (consp attributes)
 958                           (progn
 959                             (dolist (name attributes)
 960                               (unless (memq name char-db-ignored-attributes)
 961                                 (if (find-charset name)
 962                                     (push name ccss))
 963                                 (push name atr-d)))
 964                             atr-d))
 965                     (dolist (name (char-attribute-list))
 966                       (unless (memq name char-db-ignored-attributes)
 967                         (if (find-charset name)
 968                             (push name ccss))
 969                         (push name atr-d)))
 970                     atr-d)
 971                   #'char-attribute-name<)))
 972     (insert "(")
 973     (when (memq '<-subsumptive attributes)
 974       (when (or readable (not for-sub-node))
 975         (when (setq value (get-char-attribute char '<-subsumptive))
 976           (char-db-insert-relation-feature char '<-subsumptive value
 977                                            line-breaking
 978                                            ccss readable)))
 979       (setq attributes (delq '<-subsumptive attributes)))
 980     (when (and (memq '<-denotational attributes)
 981                (setq value (get-char-attribute char '<-denotational)))
 982       (char-db-insert-relation-feature char '<-denotational value
 983                                        line-breaking
 984                                        ccss readable)
 985       (setq attributes (delq '<-denotational attributes)))
 986     (when (and (memq '<-denotational@component attributes)
 987                (setq value (get-char-attribute char '<-denotational@component)))
 988       (char-db-insert-relation-feature char '<-denotational@component value
 989                                        line-breaking
 990                                        ccss readable)
 991       (setq attributes (delq '<-denotational@component attributes)))
 992     (when (and (memq '<-denotational@usage attributes)
 993                (setq value (get-char-attribute char '<-denotational@usage)))
 994       (char-db-insert-relation-feature char '<-denotational@usage value
 995                                        line-breaking
 996                                        ccss readable)
 997       (setq attributes (delq '<-denotational@usage attributes)))
 998     (when (and (memq 'name attributes)
 999                (setq value (get-char-attribute char 'name)))
1000       (insert (format
1001                (if (> (+ (current-column) (length value)) 48)
1002                    "(name . %S)%s"
1003                  "(name               . %S)%s")
1004                value line-breaking))
1005       (setq attributes (delq 'name attributes))
1006       )
1007     (when (and (memq 'name* attributes)
1008                (setq value (get-char-attribute char 'name*)))
1009       (insert (format
1010                (if (> (+ (current-column) (length value)) 48)
1011                    "(name* . %S)%s"
1012                  "(name*              . %S)%s")
1013                value line-breaking))
1014       (setq attributes (delq 'name* attributes))
1015       )
1016     (when (and (memq 'script attributes)
1017                (setq value (get-char-attribute char 'script)))
1018       (insert (format "(script\t\t%s)%s"
1019                       (mapconcat (function prin1-to-string)
1020                                  value " ")
1021                       line-breaking))
1022       (setq attributes (delq 'script attributes))
1023       )
1024     (dolist (name '(=>ucs =>ucs*))
1025       (when (and (memq name attributes)
1026                  (setq value (get-char-attribute char name)))
1027         (insert (format "(%-18s . #x%04X)\t; %c%s"
1028                         name value (decode-char '=ucs value)
1029                         line-breaking))
1030         (setq attributes (delq name attributes))))
1031     (dolist (name '(=>ucs@gb =>ucs@big5))
1032       (when (and (memq name attributes)
1033                  (setq value (get-char-attribute char name)))
1034         (insert (format "(%-18s . #x%04X)\t; %c%s"
1035                         name value
1036                         (decode-char (intern
1037                                       (concat "="
1038                                               (substring
1039                                                (symbol-name name) 2)))
1040                                      value)
1041                         line-breaking))
1042         (setq attributes (delq name attributes))
1043         ))
1044     ;; (dolist (name '(=>daikanwa))
1045     ;;   (when (and (memq name attributes)
1046     ;;              (setq value (get-char-attribute char name)))
1047     ;;     (insert
1048     ;;      (if (integerp value)
1049     ;;          (format "(%-18s . %05d)\t; %c%s"
1050     ;;                  name value (decode-char '=daikanwa value)
1051     ;;                  line-breaking)
1052     ;;        (format "(%-18s %s)\t; %c%s"
1053     ;;                name
1054     ;;                (mapconcat (function prin1-to-string)
1055     ;;                           value " ")
1056     ;;                (char-representative-of-daikanwa char)
1057     ;;                line-breaking)))
1058     ;;     (setq attributes (delq name attributes))))
1059     (when (and (memq 'general-category attributes)
1060                (setq value (get-char-attribute char 'general-category)))
1061       (insert (format
1062                "(general-category\t%s) ; %s%s"
1063                (mapconcat (lambda (cell)
1064                             (format "%S" cell))
1065                           value " ")
1066                (cond ((rassoc value unidata-normative-category-alist)
1067                       "Normative Category")
1068                      ((rassoc value unidata-informative-category-alist)
1069                       "Informative Category")
1070                      (t
1071                       "Unknown Category"))
1072                line-breaking))
1073       (setq attributes (delq 'general-category attributes))
1074       )
1075     (when (and (memq 'bidi-category attributes)
1076                (setq value (get-char-attribute char 'bidi-category)))
1077       (insert (format "(bidi-category\t. %S)%s"
1078                       value
1079                       line-breaking))
1080       (setq attributes (delq 'bidi-category attributes))
1081       )
1082     (unless (or (not (memq 'mirrored attributes))
1083                 (eq (setq value (get-char-attribute char 'mirrored 'empty))
1084                     'empty))
1085       (insert (format "(mirrored\t\t. %S)%s"
1086                       value
1087                       line-breaking))
1088       (setq attributes (delq 'mirrored attributes))
1089       )
1090     (cond
1091      ((and (memq 'decimal-digit-value attributes)
1092            (setq value (get-char-attribute char 'decimal-digit-value)))
1093       (insert (format "(decimal-digit-value . %S)%s"
1094                       value
1095                       line-breaking))
1096       (setq attributes (delq 'decimal-digit-value attributes))
1097       (when (and (memq 'digit-value attributes)
1098                  (setq value (get-char-attribute char 'digit-value)))
1099         (insert (format "(digit-value\t . %S)%s"
1100                         value
1101                         line-breaking))
1102         (setq attributes (delq 'digit-value attributes))
1103         )
1104       (when (and (memq 'numeric-value attributes)
1105                  (setq value (get-char-attribute char 'numeric-value)))
1106         (insert (format "(numeric-value\t . %S)%s"
1107                         value
1108                         line-breaking))
1109         (setq attributes (delq 'numeric-value attributes))
1110         )
1111       )
1112      (t
1113       (when (and (memq 'digit-value attributes)
1114                  (setq value (get-char-attribute char 'digit-value)))
1115         (insert (format "(digit-value\t. %S)%s"
1116                         value
1117                         line-breaking))
1118         (setq attributes (delq 'digit-value attributes))
1119         )
1120       (when (and (memq 'numeric-value attributes)
1121                  (setq value (get-char-attribute char 'numeric-value)))
1122         (insert (format "(numeric-value\t. %S)%s"
1123                         value
1124                         line-breaking))
1125         (setq attributes (delq 'numeric-value attributes))
1126         )))
1127     (when (and (memq 'iso-10646-comment attributes)
1128                (setq value (get-char-attribute char 'iso-10646-comment)))
1129       (insert (format "(iso-10646-comment\t. %S)%s"
1130                       value
1131                       line-breaking))
1132       (setq attributes (delq 'iso-10646-comment attributes))
1133       )
1134     (when (and (memq 'morohashi-daikanwa attributes)
1135                (setq value (get-char-attribute char 'morohashi-daikanwa)))
1136       (insert (format "(morohashi-daikanwa\t%s)%s"
1137                       (mapconcat (function prin1-to-string) value " ")
1138                       line-breaking))
1139       (setq attributes (delq 'morohashi-daikanwa attributes))
1140       )
1141     (setq radical nil
1142           strokes nil)
1143     (when (and (memq 'ideographic-radical attributes)
1144                (setq value (get-char-attribute char 'ideographic-radical)))
1145       (setq radical value)
1146       (insert (format "(ideographic-radical . %S)\t; %c%s"
1147                       radical
1148                       (ideographic-radical radical)
1149                       line-breaking))
1150       (setq attributes (delq 'ideographic-radical attributes))
1151       )
1152     (when (and (memq 'shuowen-radical attributes)
1153                (setq value (get-char-attribute char 'shuowen-radical)))
1154       (insert (format "(shuowen-radical\t. %S)\t; %c%s"
1155                       value
1156                       (shuowen-radical value)
1157                       line-breaking))
1158       (setq attributes (delq 'shuowen-radical attributes))
1159       )
1160     (let (key)
1161       (dolist (domain
1162                (append
1163                 char-db-feature-domains
1164                 (let (dest domain)
1165                   (dolist (feature (char-attribute-list))
1166                     (setq feature (symbol-name feature))
1167                     (when (string-match
1168                            "\\(radical\\|strokes\\)@\\([^@*]+\\)\\(\\*\\|$\\)"
1169                            feature)
1170                       (setq domain (intern (match-string 2 feature)))
1171                      (unless (memq domain dest)
1172                        (setq dest (cons domain dest)))))
1173                   (sort dest #'string<))))
1174         (setq key (intern (format "%s@%s" 'ideographic-radical domain)))
1175         (when (and (memq key attributes)
1176                    (setq value (get-char-attribute char key)))
1177           (setq radical value)
1178           (insert (format "(%s . %S)\t; %c%s"
1179                           key
1180                           radical
1181                           (ideographic-radical radical)
1182                           line-breaking))
1183           (setq attributes (delq key attributes))
1184           )
1185         (setq key (intern (format "%s@%s" 'ideographic-strokes domain)))
1186         (when (and (memq key attributes)
1187                    (setq value (get-char-attribute char key)))
1188           (setq strokes value)
1189           (insert (format "(%s . %S)%s"
1190                           key
1191                           strokes
1192                           line-breaking))
1193           (setq attributes (delq key attributes))
1194           )
1195         (setq key (intern (format "%s@%s" 'total-strokes domain)))
1196         (when (and (memq key attributes)
1197                    (setq value (get-char-attribute char key)))
1198           (insert (format "(%s       . %S)%s"
1199                           key
1200                           value
1201                           line-breaking))
1202           (setq attributes (delq key attributes))
1203           )
1204         (dolist (feature '(ideographic-radical
1205                            ideographic-strokes
1206                            total-strokes))
1207           (setq key (intern (format "%s@%s*sources" feature domain)))
1208           (when (and (memq key attributes)
1209                      (setq value (get-char-attribute char key)))
1210             (insert (format "(%s%s" key line-breaking))
1211             (dolist (cell value)
1212               (insert (format " %s" cell)))
1213             (insert ")")
1214             (insert line-breaking)
1215             (setq attributes (delq key attributes))
1216             ))
1217         ))
1218     (when (and (memq 'ideographic-strokes attributes)
1219                (setq value (get-char-attribute char 'ideographic-strokes)))
1220       (setq strokes value)
1221       (insert (format "(ideographic-strokes . %S)%s"
1222                       strokes
1223                       line-breaking))
1224       (setq attributes (delq 'ideographic-strokes attributes))
1225       )
1226     (when (and (memq 'kangxi-radical attributes)
1227                (setq value (get-char-attribute char 'kangxi-radical)))
1228       (unless (eq value radical)
1229         (insert (format "(kangxi-radical\t . %S)\t; %c%s"
1230                         value
1231                         (ideographic-radical value)
1232                         line-breaking))
1233         (or radical
1234             (setq radical value)))
1235       (setq attributes (delq 'kangxi-radical attributes))
1236       )
1237     (when (and (memq 'kangxi-strokes attributes)
1238                (setq value (get-char-attribute char 'kangxi-strokes)))
1239       (unless (eq value strokes)
1240         (insert (format "(kangxi-strokes\t . %S)%s"
1241                         value
1242                         line-breaking))
1243         (or strokes
1244             (setq strokes value)))
1245       (setq attributes (delq 'kangxi-strokes attributes))
1246       )
1247     ;; (when (and (memq 'japanese-radical attributes)
1248     ;;            (setq value (get-char-attribute char 'japanese-radical)))
1249     ;;   (unless (eq value radical)
1250     ;;     (insert (format "(japanese-radical\t . %S)\t; %c%s"
1251     ;;                     value
1252     ;;                     (ideographic-radical value)
1253     ;;                     line-breaking))
1254     ;;     (or radical
1255     ;;         (setq radical value)))
1256     ;;   (setq attributes (delq 'japanese-radical attributes))
1257     ;;   )
1258     (when (and (memq 'japanese-strokes attributes)
1259                (setq value (get-char-attribute char 'japanese-strokes)))
1260       (unless (eq value strokes)
1261         (insert (format "(japanese-strokes\t . %S)%s"
1262                         value
1263                         line-breaking))
1264         (or strokes
1265             (setq strokes value)))
1266       (setq attributes (delq 'japanese-strokes attributes))
1267       )
1268     (when (and (memq 'cns-radical attributes)
1269                (setq value (get-char-attribute char 'cns-radical)))
1270       (insert (format "(cns-radical\t . %S)\t; %c%s"
1271                       value
1272                       (ideographic-radical value)
1273                       line-breaking))
1274       (setq attributes (delq 'cns-radical attributes))
1275       )
1276     (when (and (memq 'cns-strokes attributes)
1277                (setq value (get-char-attribute char 'cns-strokes)))
1278       (unless (eq value strokes)
1279         (insert (format "(cns-strokes\t . %S)%s"
1280                         value
1281                         line-breaking))
1282         (or strokes
1283             (setq strokes value)))
1284       (setq attributes (delq 'cns-strokes attributes))
1285       )
1286     ;; (when (and (memq 'ideographic- attributes)
1287     ;;            (setq value (get-char-attribute char 'ideographic-)))
1288     ;;   (insert "(ideographic-       ")
1289     ;;   (setq lbs (concat "\n" (make-string (current-column) ?\ ))
1290     ;;         separator nil)
1291     ;;   (while (consp value)
1292     ;;     (setq cell (car value))
1293     ;;     (if (integerp cell)
1294     ;;         (setq cell (decode-char '=ucs cell)))
1295     ;;     (cond ((characterp cell)
1296     ;;            (if separator
1297     ;;                (insert lbs))
1298     ;;            (if readable
1299     ;;                (insert (format "%S" cell))
1300     ;;              (char-db-insert-char-spec cell readable))
1301     ;;            (setq separator lbs))
1302     ;;           ((consp cell)
1303     ;;            (if separator
1304     ;;                (insert lbs))
1305     ;;            (if (consp (car cell))
1306     ;;                (char-db-insert-char-spec cell readable)
1307     ;;              (char-db-insert-char-reference cell readable))
1308     ;;            (setq separator lbs))
1309     ;;           (t
1310     ;;            (if separator
1311     ;;                (insert separator))
1312     ;;            (insert (prin1-to-string cell))
1313     ;;            (setq separator " ")))
1314     ;;     (setq value (cdr value)))
1315     ;;   (insert ")")
1316     ;;   (insert line-breaking)
1317     ;;   (setq attributes (delq 'ideographic- attributes)))
1318     (when (and (memq 'total-strokes attributes)
1319                (setq value (get-char-attribute char 'total-strokes)))
1320       (insert (format "(total-strokes       . %S)%s"
1321                       value
1322                       line-breaking))
1323       (setq attributes (delq 'total-strokes attributes))
1324       )
1325     ;; (when (and (memq '->ideograph attributes)
1326     ;;            (setq value (get-char-attribute char '->ideograph)))
1327     ;;   (insert (format "(->ideograph\t%s)%s"
1328     ;;                   (mapconcat (lambda (code)
1329     ;;                                (cond ((symbolp code)
1330     ;;                                       (symbol-name code))
1331     ;;                                      ((integerp code)
1332     ;;                                       (format "#x%04X" code))
1333     ;;                                      (t
1334     ;;                                       (format "%s %S"
1335     ;;                                               line-breaking code))))
1336     ;;                              value " ")
1337     ;;                   line-breaking))
1338     ;;   (setq attributes (delq '->ideograph attributes))
1339     ;;   )
1340     ;; (when (and (memq '->decomposition attributes)
1341     ;;            (setq value (get-char-attribute char '->decomposition)))
1342     ;;   (insert (format "(->decomposition\t%s)%s"
1343     ;;                   (mapconcat (lambda (code)
1344     ;;                                (cond ((symbolp code)
1345     ;;                                       (symbol-name code))
1346     ;;                                      ((characterp code)
1347     ;;                                       (if readable
1348     ;;                                           (format "%S" code)
1349     ;;                                         (format "#x%04X"
1350     ;;                                                 (char-int code))
1351     ;;                                         ))
1352     ;;                                      ((integerp code)
1353     ;;                                       (format "#x%04X" code))
1354     ;;                                      (t
1355     ;;                                       (format "%s%S" line-breaking code))))
1356     ;;                              value " ")
1357     ;;                   line-breaking))
1358     ;;   (setq attributes (delq '->decomposition attributes))
1359     ;;   )
1360     (if (equal (get-char-attribute char '->titlecase)
1361                (get-char-attribute char '->uppercase))
1362         (setq attributes (delq '->titlecase attributes)))
1363     (when (and (memq '->mojikyo attributes)
1364                (setq value (get-char-attribute char '->mojikyo)))
1365       (insert (format "(->mojikyo\t\t. %06d)\t; %c%s"
1366                       value (decode-char 'mojikyo value)
1367                       line-breaking))
1368       (setq attributes (delq '->mojikyo attributes))
1369       )
1370     (unless readable
1371       (dolist (ignored '(composition
1372                          ->denotational <-subsumptive ->ucs-unified
1373                          ;; ->ideographic-component-forms
1374                          ))
1375         (setq attributes (delq ignored attributes))))
1376     (while attributes
1377       (setq name (car attributes))
1378       (unless (eq (setq value (get-char-attribute char name 'value-is-empty))
1379                   'value-is-empty)
1380         (cond ((setq ret (find-charset name))
1381                (setq name (charset-name ret))
1382                (when (not (memq name dest-ccss))
1383                  (setq dest-ccss (cons name dest-ccss))
1384                  (char-db-insert-ccs-feature name value line-breaking))
1385                )
1386               ((string-match "^=>ucs@" (symbol-name name))
1387                (insert (format "(%-18s . #x%04X)\t; %c%s"
1388                                name value (decode-char '=ucs value)
1389                                line-breaking))
1390                )
1391               ((eq name 'jisx0208-1978/4X)
1392                (insert (format "(%-18s . #x%04X)%s"
1393                                name value
1394                                line-breaking))
1395                )
1396               ((and
1397                 (not readable)
1398                 (not (eq name '->subsumptive))
1399                 (not (eq name '->uppercase))
1400                 (not (eq name '->lowercase))
1401                 (not (eq name '->titlecase))
1402                 (not (eq name '->canonical))
1403                 (not (eq name '->Bopomofo))
1404                 (not (eq name '->mistakable))
1405                 (not (eq name '->ideographic-variants))
1406                 (null (get-char-attribute
1407                        char (intern (format "%s*sources" name))))
1408                 (not (string-match "\\*sources$" (symbol-name name)))
1409                 (null (get-char-attribute
1410                        char (intern (format "%s*note" name))))
1411                 (not (string-match "\\*note$" (symbol-name name)))
1412                 (or (eq name '<-identical)
1413                     (eq name '<-uppercase)
1414                     (eq name '<-lowercase)
1415                     (eq name '<-titlecase)
1416                     (eq name '<-canonical)
1417                     (eq name '<-ideographic-variants)
1418                     ;; (eq name '<-synonyms)
1419                     (string-match "^<-synonyms" (symbol-name name))
1420                     (eq name '<-mistakable)
1421                     (when (string-match "^->" (symbol-name name))
1422                       (cond
1423                        ((string-match "^->fullwidth" (symbol-name name))
1424                         (not (and (consp value)
1425                                   (characterp (car value))
1426                                   (encode-char
1427                                    (car value) '=ucs 'defined-only)))
1428                         )
1429                        (t)))
1430                     ))
1431                )
1432               ((or (eq name 'ideographic-structure)
1433                    (eq name 'ideographic-structure@apparent)
1434                    (eq name 'ideographic-structure@apparent/leftmost)
1435                    (eq name 'ideographic-combination)
1436                    ;; (eq name 'ideographic-)
1437                    (eq name '=decomposition)
1438                    (char-feature-base-name= '=decomposition name)
1439                    (char-feature-base-name= '=>decomposition name)
1440                    ;; (string-match "^=>*decomposition\\(@[^*]+\\)?$"
1441                    ;;               (symbol-name name))
1442                    (string-match "^\\(->\\|<-\\)[^*]*$" (symbol-name name))
1443                    (string-match "^\\(->\\|<-\\)[^*]*\\*sources$"
1444                                  (symbol-name name))
1445                    )
1446                (char-db-insert-relation-feature char name value
1447                                                 line-breaking
1448                                                 ccss readable))
1449               ;; ((memq name '(ideograph=
1450               ;;               original-ideograph-of
1451               ;;               ancient-ideograph-of
1452               ;;               vulgar-ideograph-of
1453               ;;               wrong-ideograph-of
1454               ;;               ;; simplified-ideograph-of
1455               ;;               ideographic-variants
1456               ;;               ;; ideographic-different-form-of
1457               ;;               ))
1458               ;;  (insert (format "(%-18s%s " name line-breaking))
1459               ;;  (setq lbs (concat "\n" (make-string (current-column) ?\ ))
1460               ;;        separator nil)
1461               ;;  (while (consp value)
1462               ;;    (setq cell (car value))
1463               ;;    (if (and (consp cell)
1464               ;;             (consp (car cell)))
1465               ;;        (progn
1466               ;;          (if separator
1467               ;;              (insert lbs))
1468               ;;          (char-db-insert-alist cell readable)
1469               ;;          (setq separator lbs))
1470               ;;      (if separator
1471               ;;          (insert separator))
1472               ;;      (insert (prin1-to-string cell))
1473               ;;      (setq separator " "))
1474               ;;    (setq value (cdr value)))
1475               ;;  (insert ")")
1476               ;;  (insert line-breaking))
1477               ((consp value)
1478                (insert (format "(%-18s " name))
1479                (setq lbs (concat "\n" (make-string (current-column) ?\ ))
1480                      separator nil)
1481                (while (consp value)
1482                  (setq cell (car value))
1483                  (if (and (consp cell)
1484                           (consp (car cell))
1485                           (setq ret (condition-case nil
1486                                         (find-char cell)
1487                                       (error nil))))
1488                      (progn
1489                        (setq rest cell
1490                              al nil
1491                              cal nil)
1492                        (while rest
1493                          (setq key (car (car rest)))
1494                          (if (find-charset key)
1495                              (setq cal (cons key cal))
1496                            (setq al (cons key al)))
1497                          (setq rest (cdr rest)))
1498                        (if separator
1499                            (insert lbs))
1500                        (insert-char-attributes ret
1501                                                readable
1502                                                al ; cal
1503                                                nil 'for-sub-node)
1504                        (setq separator lbs))
1505                    (setq ret (prin1-to-string cell))
1506                    (if separator
1507                        (if (< (+ (current-column)
1508                                  (length ret)
1509                                  (length separator))
1510                               76)
1511                            (insert separator)
1512                          (insert lbs)))
1513                    (insert ret)
1514                    (setq separator " "))
1515                  (setq value (cdr value)))
1516                (insert ")")
1517                (insert line-breaking))
1518               (t
1519                (insert (format "(%-18s" name))
1520                (setq ret (prin1-to-string value))
1521                (unless (< (+ (current-column)
1522                              (length ret)
1523                              3)
1524                           76)
1525                  (insert line-breaking))
1526                (insert " . " ret ")" line-breaking)
1527                ;; (insert (format "(%-18s . %S)%s"
1528                ;;                 name value
1529                ;;                 line-breaking))
1530                )
1531               ))
1532       (setq attributes (cdr attributes)))
1533     (insert ")")))
1534
1535 (defun insert-char-data (char &optional readable
1536                               attributes)
1537   (save-restriction
1538     (narrow-to-region (point)(point))
1539     (insert "(define-char
1540   '")
1541     (insert-char-attributes char readable attributes)
1542     (insert ")\n")
1543     (goto-char (point-min))
1544     (while (re-search-forward "[ \t]+$" nil t)
1545       (replace-match ""))
1546     ;; from tabify.
1547     (goto-char (point-min))
1548     (while (re-search-forward "[ \t][ \t][ \t]*" nil t)
1549       (let ((column (current-column))
1550             (indent-tabs-mode t))
1551         (delete-region (match-beginning 0) (point))
1552         (indent-to column)))
1553     (goto-char (point-max))
1554     ;; (tabify (point-min)(point-max))
1555     ))
1556
1557 (defun insert-char-data-with-variant (char &optional printable
1558                                            no-ucs-unified
1559                                            script excluded-script)
1560   (insert-char-data char printable)
1561   (let ((variants (char-variants char))
1562         rest
1563         variant vs ret)
1564     (setq variants (sort variants #'<))
1565     (setq rest variants)
1566     (setq variants (cons char variants))
1567     (while rest
1568       (setq variant (car rest))
1569       (unless (get-char-attribute variant '<-subsumptive)
1570         (if (and (or (null script)
1571                      (null (setq vs (get-char-attribute variant 'script)))
1572                      (memq script vs))
1573                  (or (null excluded-script)
1574                      (null (setq vs (get-char-attribute variant 'script)))
1575                      (not (memq excluded-script vs))))
1576             (unless (and no-ucs-unified (get-char-attribute variant '=ucs))
1577               (insert-char-data variant printable)
1578               (if (setq ret (char-variants variant))
1579                   (while ret
1580                     (or (memq (car ret) variants)
1581                         ;; (get-char-attribute (car ret) '<-subsumptive)
1582                         (setq rest (nconc rest (list (car ret)))))
1583                     (setq ret (cdr ret)))))))
1584       (setq rest (cdr rest)))))
1585
1586 (defun insert-char-range-data (min max &optional script excluded-script)
1587   (let ((code min)
1588         char)
1589     (while (<= code max)
1590       (setq char (decode-char '=ucs code))
1591       (if (encode-char char '=ucs 'defined-only)
1592           (insert-char-data-with-variant char nil 'no-ucs-unified
1593                                          script excluded-script))
1594       (setq code (1+ code)))))
1595
1596 (defun write-char-range-data-to-file (min max file
1597                                           &optional script excluded-script)
1598   (let ((coding-system-for-write char-db-file-coding-system))
1599     (with-temp-buffer
1600       (insert (format ";; -*- coding: %s -*-\n"
1601                       char-db-file-coding-system))
1602       (insert-char-range-data min max script excluded-script)
1603       (write-region (point-min)(point-max) file))))
1604
1605 (defvar what-character-original-window-configuration)
1606
1607 ;;;###autoload
1608 (defun what-char-definition (char)
1609   (interactive (list (char-after)))
1610   (let ((buf (get-buffer-create "*Character Description*"))
1611         (the-buf (current-buffer))
1612         (win-conf (current-window-configuration)))
1613     (pop-to-buffer buf)
1614     (make-local-variable 'what-character-original-window-configuration)
1615     (setq what-character-original-window-configuration win-conf)
1616     (setq buffer-read-only nil)
1617     (erase-buffer)
1618     (condition-case err
1619         (progn
1620           (insert-char-data-with-variant char 'printable)
1621           (unless (char-attribute-alist char)
1622             (insert (format ";; = %c\n"
1623                             (let* ((rest (split-char char))
1624                                    (ccs (pop rest))
1625                                    (code (pop rest)))
1626                               (while rest
1627                                 (setq code (logior (lsh code 8)
1628                                                    (pop rest))))
1629                               (decode-char ccs code)))))
1630           ;; (char-db-update-comment)
1631           (set-buffer-modified-p nil)
1632           (view-mode the-buf (lambda (buf)
1633                                (set-window-configuration
1634                                 what-character-original-window-configuration)
1635                                ))
1636           (goto-char (point-min)))
1637       (error (progn
1638                (set-window-configuration
1639                 what-character-original-window-configuration)
1640                (signal (car err) (cdr err)))))))
1641
1642
1643 ;;; @ end
1644 ;;;
1645
1646 (provide 'char-db-util)
1647
1648 ;;; char-db-util.el ends here