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