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