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