git.chise.org Git - chise/xemacs-chise.git.1/blob - 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 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 ?谷 ?只 ?㕯 ; 050
  72    ?句 ?丩 ?古 ?十 ?卅 ?言 ?誩 ?音 ?䇂 ?丵 ; 060
  73    ?菐 ?𠬞 ?𠬜 ?共 ?異 ?舁 ?𦥑 ?䢅 ?爨 ?革 ; 070
  74    ?鬲 ?䰜 ?爪 ?𠃨 ?鬥 ?又 ?𠂇 ?㕜 ?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    ?萑 ?𦫳 ?苜 ?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 ?丰 ; 140
  81    ?耒 ?𧢲 ?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 ?卣 ?齊 ?朿 ?片 ?鼎 ; 250
  93    ?克 ?彔 ?禾 ?秝 ?黍 ?香 ?米 ?毇 ?臼 ?凶 ; 260
  94    ?𣎳 ?林 ?麻 ?尗 ?耑 ?韭 ?瓜 ?瓠 ?宀 ?宮 ; 270
  95    ?呂 ?穴 ?㝱 ?𤕫 ?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    ])
 108
 109 (defun shuowen-radical (number)
 110   (aref shuowen-radicals (1- number)))
 111
 112 (defvar char-db-file-coding-system 'utf-8-mcs-er)
 113
 114 (defvar char-db-ignored-attributes '(ideographic-products))
 115
 116 (defvar char-db-coded-charset-priority-list
 117   '(ascii
 118     control-1
 119     latin-iso8859-1
 120     latin-iso8859-2
 121     latin-iso8859-3
 122     latin-iso8859-4
 123     latin-iso8859-9
 124     latin-jisx0201
 125     cyrillic-iso8859-5
 126     greek-iso8859-7
 127     thai-tis620
 128     ;; =mj
 129     ;; =adobe-japan1-0
 130     ;; =adobe-japan1-1
 131     ;; =adobe-japan1-2
 132     ;; =adobe-japan1-3
 133     ;; =adobe-japan1-4
 134     ;; =adobe-japan1-5
 135     ;; =adobe-japan1-6
 136     =jis-x0208
 137     =jis-x0208@1978
 138     =jis-x0208@1983
 139     japanese-jisx0212
 140     chinese-gb2312
 141     =jis-x0208@1990
 142     chinese-cns11643-1
 143     chinese-cns11643-2
 144     chinese-cns11643-3
 145     chinese-cns11643-4
 146     chinese-cns11643-5
 147     chinese-cns11643-6
 148     chinese-cns11643-7
 149     =jis-x0213-1
 150     =jis-x0213-1@2000
 151     =jis-x0213-1@2004
 152     =jis-x0213-2
 153     korean-ksc5601
 154     chinese-isoir165
 155     katakana-jisx0201
 156     hebrew-iso8859-8
 157     chinese-gb12345
 158     latin-viscii
 159     ethiopic-ucs
 160     =big5-cdp
 161     =hanyo-denshi/ja
 162     =hanyo-denshi/jb
 163     =hanyo-denshi/jc
 164     =hanyo-denshi/jd
 165     =hanyo-denshi/ft
 166     =hanyo-denshi/ia
 167     =hanyo-denshi/ib
 168     =hanyo-denshi/hg
 169     =hanyo-denshi/jt
 170     =hanyo-denshi/ks
 171     =hanyo-denshi/tk
 172     ;; ==mj
 173     ;; ==adobe-japan1-0
 174     ;; ==adobe-japan1-1
 175     ;; ==adobe-japan1-2
 176     ;; ==adobe-japan1-3
 177     ;; ==adobe-japan1-4
 178     ;; ==adobe-japan1-5
 179     ;; ==adobe-japan1-6
 180     ==jis-x0208
 181     ==jis-x0213-1
 182     ==jis-x0213-2
 183     ==jis-x0212
 184     ==hanyo-denshi/ja
 185     ==hanyo-denshi/jb
 186     ==hanyo-denshi/jc
 187     ==hanyo-denshi/ft
 188     ==hanyo-denshi/ia
 189     ==hanyo-denshi/ib
 190     ==hanyo-denshi/hg
 191     ==hanyo-denshi/jt
 192     ==hanyo-denshi/ks
 193     ==hanyo-denshi/tk
 194     =gt
 195     =gt-k
 196     =daikanwa
 197     =daikanwa@rev2
 198     =daikanwa@rev1
 199     =daikanwa/+p
 200     ==daikanwa
 201     =cbeta
 202     ideograph-hanziku-1
 203     ideograph-hanziku-2
 204     ideograph-hanziku-3
 205     ideograph-hanziku-4
 206     ideograph-hanziku-5
 207     ideograph-hanziku-6
 208     ideograph-hanziku-7
 209     ideograph-hanziku-8
 210     ideograph-hanziku-9
 211     ideograph-hanziku-10
 212     ideograph-hanziku-11
 213     ideograph-hanziku-12
 214     ;; =>>>adobe-japan1-0
 215     ;; =>>>adobe-japan1-1
 216     ;; =>>>adobe-japan1-2
 217     ;; =>>>adobe-japan1-3
 218     ;; =>>>adobe-japan1-4
 219     ;; =>>>adobe-japan1-5
 220     ;; =>>>adobe-japan1-6
 221     ;; =>>>jis-x0208
 222     ;; =>>>jis-x0213-1
 223     ;; =>>>jis-x0213-2
 224     ;; =>>>hanyo-denshi/ja
 225     ;; =>>>hanyo-denshi/jb
 226     ;; =>>>hanyo-denshi/jc
 227     ;; =>>>hanyo-denshi/ft
 228     ;; =>>>hanyo-denshi/ib
 229     ;; =>>>hanyo-denshi/hg
 230     ;; =>>>hanyo-denshi/jt
 231     ;; =>>>hanyo-denshi/ks
 232     ;; =>>>gt
 233     =>>adobe-japan1-0
 234     =>>adobe-japan1-1
 235     =>>adobe-japan1-2
 236     =>>adobe-japan1-3
 237     =>>adobe-japan1-4
 238     =>>adobe-japan1-5
 239     =>>adobe-japan1-6
 240     =>>jis-x0208
 241     =>>jis-x0213-1
 242     =>>jis-x0213-1@2000
 243     =>>jis-x0213-1@2004
 244     =>>jis-x0213-2
 245     =>>jis-x0208@1978
 246     =>>hanyo-denshi/ft
 247     =>>hanyo-denshi/jt
 248     =>>hanyo-denshi/ks
 249     =>>gt
 250     =>>daikanwa
 251     =>>cbeta
 252     =+>jis-x0208
 253     =+>jis-x0213-1
 254     =+>jis-x0213-2
 255     =+>adobe-japan1-0
 256     =+>adobe-japan1-1
 257     =+>adobe-japan1-2
 258     =+>adobe-japan1-3
 259     =+>adobe-japan1-4
 260     =+>adobe-japan1-5
 261     =+>adobe-japan1-6
 262     =+>jis-x0208@1978
 263     =>jis-x0208
 264     =>jis-x0208@1997
 265     =>jis-x0213-1
 266     =>jis-x0213-1@2000
 267     =>jis-x0213-1@2004
 268     =>jis-x0213-2
 269     ==>ucs@bucs
 270     =>iwds-1
 271     ;; =>ucs@hanyo-denshi
 272     =>ucs@iso
 273     =>ucs@unicode
 274     =>ucs@jis
 275     =>ucs@cns
 276     =>ucs@ks
 277     =+>ucs@iso
 278     =+>ucs@unicode
 279     =+>ucs@jis
 280     =+>ucs@jis/1990
 281     =+>ucs@cns
 282     =+>ucs@ks
 283     =>>ucs@iso
 284     =>>ucs@unicode
 285     =>>ucs@jis
 286     =>>ucs@cns
 287     =>>>ucs@iso
 288     =>>>ucs@unicode
 289     ==ucs@iso
 290     ==ucs@unicode
 291     ;; ==ucs@cns
 292     ==gb2312
 293     ==ks-x1001
 294     ==cns11643-1
 295     ==cns11643-2
 296     ==cns11643-3
 297     ==cns11643-4
 298     ==cns11643-5
 299     ==cns11643-6
 300     ==cns11643-7
 301     ==gt
 302     ==jis-x0208@1990
 303     ;; ==jis-x0208@1983
 304     ==jis-x0208@1978
 305     ==gt-k
 306     =ucs@iso
 307     =ucs@unicode
 308     =ucs@cns
 309     ==big5-cdp
 310     ==cbeta
 311     =>>big5-cdp
 312     =>>gt-k
 313     =+>gt
 314     =>gt
 315     =>mj
 316     =>big5-cdp
 317     =>daikanwa
 318     =>daikanwa/ho
 319     =>cns11643-5
 320     =>cns11643-7
 321     =big5
 322     =big5-eten
 323     =>gt-k
 324     =zinbun-oracle
 325     =>zinbun-oracle
 326     =ruimoku-v6
 327     =>>ruimoku-v6
 328     ==ruimoku-v6
 329     =jef-china3
 330     =>cbeta
 331     =shinjigen
 332     =ucs-var-001
 333     =ucs-var-002
 334     =ucs-var-003
 335     =ucs-var-004
 336     =ucs-var-005
 337     =ucs-var-006
 338     =ucs-var-008
 339     =ucs-var-010
 340     =ucs-itaiji-001
 341     =ucs-itaiji-002
 342     =ucs-itaiji-003
 343     =ucs-itaiji-004
 344     =ucs-itaiji-005
 345     =ucs-itaiji-006
 346     =ucs-itaiji-007
 347     =ucs-itaiji-008
 348     =ucs-itaiji-009
 349     =ucs-itaiji-010
 350     =ucs-itaiji-011
 351     =>ucs-itaiji-001
 352     =>ucs-itaiji-002
 353     =>ucs-itaiji-003
 354     =>ucs-itaiji-004
 355     =>ucs-itaiji-005
 356     =>ucs-itaiji-006
 357     =>ucs-itaiji-007
 358     =>ucs-itaiji-009
 359     =big5-cdp-var-001
 360     =big5-cdp-var-002
 361     =big5-cdp-var-003
 362     =big5-cdp-var-004
 363     =big5-cdp-var-005
 364     =big5-cdp-itaiji-001
 365     =big5-cdp-itaiji-002
 366     =big5-cdp-itaiji-003
 367     =>big5-cdp-itaiji-001
 368     =>ucs@iwds-1
 369     =>ucs@cognate
 370     =>ucs@component
 371     ==ucs@gb
 372     =ucs@gb
 373     ==ucs-var-002
 374     =ucs@JP/hanazono))
 375
 376
 377 ;;; @ char-db formatters
 378 ;;;
 379
 380 (defun char-db-make-char-spec (char)
 381   (let (ret char-spec)
 382     (cond ((characterp char)
 383            (cond ((and (setq ret (encode-char char '=ucs 'defined-only))
 384                        (not (and (<= #xE000 ret)(<= ret #xF8FF))))
 385                   (setq char-spec (list (cons '=ucs ret)))
 386                   (cond ((setq ret (get-char-attribute char 'name))
 387                          (setq char-spec (cons (cons 'name ret) char-spec))
 388                          )
 389                         ((setq ret (get-char-attribute char 'name*))
 390                          (setq char-spec (cons (cons 'name* ret) char-spec))
 391                          ))
 392                   )
 393                  ((encode-char char '=mj 'defined-only)
 394                   (setq char-spec nil)
 395                   (dolist (ccs (charset-list))
 396                     (if (and (or (eq ccs '=mj)
 397                                  ;; (eq (charset-property ccs 'iso-ir) 177)
 398                                  (string-match "=ucs@" (symbol-name ccs))
 399                                  )
 400                              (setq ccs (charset-name ccs))
 401                              (null (assq ccs char-spec))
 402                              (setq ret (encode-char char ccs 'defined-only)))
 403                         (setq char-spec (cons (cons ccs ret) char-spec))))
 404                   )
 405                  ((encode-char char '==mj 'defined-only)
 406                   (setq char-spec nil)
 407                   (dolist (ccs (charset-list))
 408                     (if (and (or (eq ccs '==mj)
 409                                  ;; (eq (charset-property ccs 'iso-ir) 177)
 410                                  (string-match "=ucs@" (symbol-name ccs))
 411                                  )
 412                              (setq ccs (charset-name ccs))
 413                              (null (assq ccs char-spec))
 414                              (setq ret (encode-char char ccs 'defined-only)))
 415                         (setq char-spec (cons (cons ccs ret) char-spec))))
 416                   )
 417                  ((encode-char char '=adobe-japan1 'defined-only)
 418                   (setq char-spec nil)
 419                   (dolist (ccs (charset-list))
 420                     (if (and (or (memq ccs
 421                                        '(=adobe-japan1-0
 422                                          =adobe-japan1-1
 423                                          =adobe-japan1-2
 424                                          =adobe-japan1-3
 425                                          =adobe-japan1-4
 426                                          =adobe-japan1-5
 427                                          =adobe-japan1-6
 428                                          ))
 429                                  ;; (eq (charset-property ccs 'iso-ir) 177)
 430                                  (string-match "=ucs@" (symbol-name ccs))
 431                                  )
 432                              (setq ccs (charset-name ccs))
 433                              (null (assq ccs char-spec))
 434                              (setq ret (encode-char char ccs 'defined-only)))
 435                         (setq char-spec (cons (cons ccs ret) char-spec))))
 436                   )
 437                  ((encode-char char '==adobe-japan1 'defined-only)
 438                   (setq char-spec nil)
 439                   (dolist (ccs (charset-list))
 440                     (if (and (or (memq ccs
 441                                        '(==adobe-japan1-0
 442                                          ==adobe-japan1-1
 443                                          ==adobe-japan1-2
 444                                          ==adobe-japan1-3
 445                                          ==adobe-japan1-4
 446                                          ==adobe-japan1-5
 447                                          ==adobe-japan1-6
 448                                          ))
 449                                  ;; (eq (charset-property ccs 'iso-ir) 177)
 450                                  (string-match "=ucs@" (symbol-name ccs))
 451                                  )
 452                              (setq ccs (charset-name ccs))
 453                              (null (assq ccs char-spec))
 454                              (setq ret (encode-char char ccs 'defined-only)))
 455                         (setq char-spec (cons (cons ccs ret) char-spec))))
 456                   )
 457                  ((setq ret
 458                         (catch 'tag
 459                           (let ((rest char-db-coded-charset-priority-list)
 460                                 ccs)
 461                             (while rest
 462                               (setq ccs (charset-name
 463                                          (find-charset (car rest))))
 464                               (if (setq ret
 465                                         (encode-char char ccs
 466                                                      'defined-only))
 467                                   (throw 'tag (cons ccs ret)))
 468                               (setq rest (cdr rest))))))
 469                   (setq char-spec (list ret))
 470                   (dolist (ccs (delq (car ret) (charset-list)))
 471                     (if (and (or (charset-iso-final-char ccs)
 472                                  (memq ccs
 473                                        '(=daikanwa
 474                                          =daikanwa@rev2
 475                                          ;; =gt-k
 476                                          =jis-x0208@1997
 477                                          ))
 478                                  (and (eq (charset-property ccs 'iso-ir) 177)
 479                                       (not (eq (charset-name ccs) '=ucs@big5)))
 480                                  ;; (string-match "=ucs@" (symbol-name ccs))
 481                                  )
 482                              (setq ccs (charset-name ccs))
 483                              (null (assq ccs char-spec))
 484                              (setq ret (encode-char char ccs 'defined-only)))
 485                         (setq char-spec (cons (cons ccs ret) char-spec))))
 486                   (if (null char-spec)
 487                       (setq char-spec (split-char char)))
 488                   (cond ((setq ret (get-char-attribute char 'name))
 489                          (setq char-spec (cons (cons 'name ret) char-spec))
 490                          )
 491                         ((setq ret (get-char-attribute char 'name*))
 492                          (setq char-spec (cons (cons 'name* ret) char-spec))
 493                          ))
 494                   )
 495                  ((setq ret (get-char-attribute
 496                              char 'ideographic-combination))
 497                   (setq char-spec
 498                         (cons (cons 'ideographic-combination ret)
 499                               char-spec))
 500                   ))
 501            char-spec)
 502           ((consp char)
 503            char))))
 504
 505 (defun char-db-insert-char-spec (char &optional readable column
 506                                       required-features)
 507   (unless column
 508     (setq column (current-column)))
 509   (let (char-spec temp-char)
 510     (setq char-spec (char-db-make-char-spec char))
 511     (unless (or (characterp char) ; char
 512                 (condition-case nil
 513                     (setq char (find-char char-spec))
 514                   (error nil)))
 515       ;; define temporary character
 516       ;;   Current implementation is dirty.
 517       (setq temp-char (define-char (cons '(ideograph-daikanwa . 0)
 518                                          char-spec)))
 519       (remove-char-attribute temp-char 'ideograph-daikanwa)
 520       (setq char temp-char))
 521     (insert-char-attributes char
 522                             readable
 523                             (union (mapcar #'car char-spec)
 524                                    required-features)
 525                             nil 'for-sub-node)
 526     (when temp-char
 527       ;; undefine temporary character
 528       ;;   Current implementation is dirty.
 529       (setq char-spec (char-attribute-alist temp-char))
 530       (while char-spec
 531         (remove-char-attribute temp-char (car (car char-spec)))
 532         (setq char-spec (cdr char-spec))))))
 533
 534 (defun char-db-insert-alist (alist &optional readable column)
 535   (unless column
 536     (setq column (current-column)))
 537   (let ((line-breaking
 538          (concat "\n" (make-string (1+ column) ?\ )))
 539         name value
 540         ret al ; cal
 541         key
 542         lbs cell rest separator)
 543     (insert "(")
 544     (while alist
 545       (setq name (car (car alist))
 546             value (cdr (car alist)))
 547       (cond ((eq name 'char)
 548              (insert "(char . ")
 549              (if (setq ret (condition-case nil
 550                                (find-char value)
 551                              (error nil)))
 552                  (progn
 553                    (setq al nil
 554                          ;; cal nil
 555                          )
 556                    (while value
 557                      (setq key (car (car value)))
 558                      ;; (if (find-charset key)
 559                      ;;     (setq cal (cons key cal))
 560                      (setq al (cons key al))
 561                      ;; )
 562                      (setq value (cdr value)))
 563                    (insert-char-attributes ret
 564                                            readable
 565                                            (or al 'none) ; cal
 566                                            nil 'for-sub-node))
 567                (insert (prin1-to-string value)))
 568              (insert ")")
 569              (insert line-breaking))
 570             ((consp value)
 571              (insert (format "(%-18s " name))
 572              (setq lbs (concat "\n" (make-string (current-column) ?\ )))
 573              (while (consp value)
 574                (setq cell (car value))
 575                (if (and (consp cell)
 576                         (consp (car cell))
 577                         (setq ret (condition-case nil
 578                                       (find-char cell)
 579                                     (error nil)))
 580                         )
 581                    (progn
 582                      (setq rest cell
 583                            al nil
 584                            ;; cal nil
 585                            )
 586                      (while rest
 587                        (setq key (car (car rest)))
 588                        ;; (if (find-charset key)
 589                        ;;     (setq cal (cons key cal))
 590                        (setq al (cons key al))
 591                        ;; )
 592                        (setq rest (cdr rest)))
 593                      (if separator
 594                          (insert lbs))
 595                      (insert-char-attributes ret
 596                                              readable
 597                                              al ; cal
 598                                              nil 'for-sub-node)
 599                      (setq separator lbs))
 600                  (if separator
 601                      (insert separator))
 602                  (insert (prin1-to-string cell))
 603                  (setq separator " "))
 604                (setq value (cdr value)))
 605              (insert ")")
 606              (insert line-breaking))
 607             (t
 608              (insert (format "(%-18s . %S)%s"
 609                              name value
 610                              line-breaking))))
 611       (setq alist (cdr alist))))
 612   (insert ")"))
 613
 614 (defun char-db-insert-char-reference (plist &optional readable column)
 615   (unless column
 616     (setq column (current-column)))
 617   (let ((line-breaking
 618          (concat "\n" (make-string (1+ column) ?\ )))
 619         (separator "")
 620         name value)
 621     (insert "(")
 622     (while plist
 623       (setq name (pop plist))
 624       (setq value (pop plist))
 625       (cond ((eq name :char)
 626              (insert separator)
 627              (insert ":char\t")
 628              (cond ((numberp value)
 629                     (setq value (decode-char '=ucs value)))
 630                    ;; ((consp value)
 631                    ;;  (setq value (or (find-char value)
 632                    ;;                  value)))
 633                    )
 634              (char-db-insert-char-spec value readable)
 635              (insert line-breaking)
 636              (setq separator ""))
 637             ((eq name :radical)
 638              (insert (format "%s%s\t%d ; %c%s"
 639                              separator
 640                              name value
 641                              (ideographic-radical value)
 642                              line-breaking))
 643              (setq separator ""))
 644             (t
 645              (insert (format "%s%s\t%S" separator name value))
 646              (setq separator line-breaking)))
 647       ))
 648   (insert ")"))
 649
 650 (defun char-db-decode-isolated-char (ccs code-point)
 651   (let (ret)
 652     (setq ret
 653           (cond ((eq ccs 'arabic-iso8859-6)
 654                  (decode-char ccs code-point))
 655                 ;; ((eq ccs '=gt)
 656                 ;;  (decode-builtin-char '==gt code-point))
 657                 ((and (memq ccs '(=gt-pj-1
 658                                   =gt-pj-2
 659                                   =gt-pj-3
 660                                   =gt-pj-4
 661                                   =gt-pj-5
 662                                   =gt-pj-6
 663                                   =gt-pj-7
 664                                   =gt-pj-8
 665                                   =gt-pj-9
 666                                   =gt-pj-10
 667                                   =gt-pj-11))
 668                       (setq ret (decode-char ccs code-point))
 669                       (setq ret (encode-char ret '=gt 'defined-only)))
 670                  (decode-builtin-char '=gt ret))
 671                 (t
 672                  (decode-builtin-char ccs code-point))))
 673     (cond ((null ret)
 674            (or (decode-char ccs code-point)
 675                (define-char (list (cons ccs code-point)))))
 676           ((and (<= 0 (char-int ret))
 677                 (<= (char-int ret) #x1F))
 678            (decode-char '=ucs (+ #x2400 (char-int ret))))
 679           ((= (char-int ret) #x7F)
 680            ?\u2421)
 681           (t ret))))
 682
 683 (defvar char-db-convert-obsolete-format t)
 684
 685 (defun char-db-insert-ccs-feature (name value line-breaking)
 686   (cond
 687    ((integerp value)
 688     (insert
 689      (format
 690       (cond
 691        ((memq name '(=>iwds-1
 692                      ==shinjigen
 693                      =shinjigen
 694                      =shinjigen@1ed ==shinjigen@1ed
 695                      =shinjigen@rev ==shinjigen@rev
 696                      =shinjigen/+p@rev ==shinjigen/+p@rev
 697                      ===daikanwa/ho ==daikanwa/ho
 698                      =daikanwa/ho =>>daikanwa/ho =>daikanwa/ho))
 699         "(%-18s .  %04d)\t; %c")
 700        ((eq name '=shinjigen@1ed/24pr)
 701         "(%-18s . %04d)\t; %c")
 702        ((or
 703          (memq name
 704                '(===daikanwa
 705                  ==daikanwa =daikanwa =>>daikanwa =>daikanwa
 706                  =daikanwa@rev1 =daikanwa@rev2
 707                  =daikanwa/+p ==daikanwa/+p ===daikanwa/+p
 708                  =>>daikanwa/+p
 709                  =daikanwa/+2p =>>daikanwa/+2p
 710                  =gt ==gt ===gt
 711                  =>>gt =+>gt =>gt
 712                  =gt-k ==gt-k ===gt-k
 713                  =>>gt-k =>gt-k
 714                  =adobe-japan1-0 ==adobe-japan1-0 ===adobe-japan1-0
 715                  =adobe-japan1-1 ==adobe-japan1-1 ===adobe-japan1-1
 716                  =adobe-japan1-2 ==adobe-japan1-2 ===adobe-japan1-2
 717                  =adobe-japan1-3 ==adobe-japan1-3 ===adobe-japan1-3
 718                  =adobe-japan1-4 ==adobe-japan1-4 ===adobe-japan1-4
 719                  =adobe-japan1-5 ==adobe-japan1-5 ===adobe-japan1-5
 720                  =adobe-japan1-6 ==adobe-japan1-6 ===adobe-japan1-6
 721                  =>>adobe-japan1-0 =+>adobe-japan1-0
 722                  =>>adobe-japan1-1 =+>adobe-japan1-1
 723                  =>>adobe-japan1-2 =+>adobe-japan1-2
 724                  =>>adobe-japan1-3 =+>adobe-japan1-3
 725                  =>>adobe-japan1-4 =+>adobe-japan1-4
 726                  =>>adobe-japan1-5 =+>adobe-japan1-5
 727                  =>>adobe-japan1-6 =+>adobe-japan1-6
 728                  =>cbeta =cbeta =>>cbeta ==cbeta ===cbeta
 729                  =zinbun-oracle =>zinbun-oracle
 730                  ===hng-jou ===hng-keg ===hng-dng ===hng-mam
 731                  ===hng-drt ===hng-kgk ===hng-myz ===hng-kda
 732                  ===hng-khi ===hng-khm ===hng-hok ===hng-kyd ===hng-sok
 733                  ===hng-yhk ===hng-kak ===hng-kar ===hng-kae
 734                  ===hng-sys ===hng-tsu ===hng-tzj
 735                  ===hng-hos ===hng-nak ===hng-jhk
 736                  ===hng-hod ===hng-gok ===hng-ink ===hng-nto
 737                  ===hng-nkm ===hng-k24 ===hng-nkk
 738                  ===hng-kcc ===hng-kcj ===hng-kbk ===hng-sik
 739                  ===hng-skk ===hng-kyu ===hng-ksk ===hng-wan
 740                  ===hng-okd ===hng-wad ===hng-kmi
 741                  ===hng-zkd ===hng-doh ===hng-jyu
 742                  ===hng-tzs ===hng-kss ===hng-kyo
 743                  ===hng-smk))
 744          ;; (string-match "^=adobe-" (symbol-name name))
 745          )
 746         "(%-18s . %05d)\t; %c")
 747        ((memq name '(=hanyo-denshi/ks
 748                      ==hanyo-denshi/ks ===hanyo-denshi/ks
 749                      =>>hanyo-denshi/ks
 750                      =koseki ==koseki
 751                      =mj ==mj ===mj =>>mj =>mj
 752                      =zihai mojikyo))
 753         "(%-18s . %06d)\t; %c")
 754        ((memq name '(=hanyo-denshi/tk ==hanyo-denshi/tk))
 755         "(%-18s . %08d)\t; %c")
 756        ((>= (charset-dimension name) 2)
 757         "(%-18s . #x%04X)\t; %c")
 758        (t
 759         "(%-18s . #x%02X)\t; %c"))
 760       name
 761       (if (= (charset-iso-graphic-plane name) 1)
 762           (logior value
 763                   (cond ((= (charset-dimension name) 1)
 764                          #x80)
 765                         ((= (charset-dimension name) 2)
 766                          #x8080)
 767                         ((= (charset-dimension name) 3)
 768                          #x808080)
 769                         (t 0)))
 770         value)
 771       (char-db-decode-isolated-char name value)))
 772     (if (and (= (charset-chars name) 94)
 773              (= (charset-dimension name) 2))
 774         (insert (format " [%02d-%02d]"
 775                         (- (lsh value -8) 32)
 776                         (- (logand value 255) 32))))
 777     )
 778    (t
 779     (insert (format "(%-18s . %s)" name value))
 780     ))
 781   (insert line-breaking))
 782
 783 (defun char-db-insert-relation-feature (char name value line-breaking
 784                                              ccss readable)
 785   (insert (format "(%-18s%s " name line-breaking))
 786   (let ((lbs (concat "\n" (make-string (current-column) ?\ )))
 787         separator cell sources required-features
 788         ret)
 789     (while (consp value)
 790       (setq cell (car value))
 791       (if (integerp cell)
 792           (setq cell (decode-char '=ucs cell)))
 793       (cond
 794        ((eq name '->subsumptive)
 795         (when (or (not (some (lambda (atr)
 796                                (get-char-attribute cell atr))
 797                              char-db-ignored-attributes))
 798                   (some (lambda (ccs)
 799                           (encode-char cell ccs 'defined-only))
 800                         ccss))
 801           (if separator
 802               (insert lbs))
 803           (let ((char-db-ignored-attributes
 804                  (cons '<-subsumptive
 805                        char-db-ignored-attributes)))
 806             (insert-char-attributes cell readable nil nil 'for-sub-node))
 807           (setq separator lbs))
 808         )
 809        ((characterp cell)
 810         (setq sources
 811               (get-char-attribute
 812                char (intern (format "%s*sources" name))))
 813         (setq required-features nil)
 814         (dolist (source sources)
 815           (cond
 816            ((memq source '(JP
 817                            JP/Jouyou
 818                            shinjigen shinjigen@1ed shinjigen@rev))
 819             (setq required-features
 820                   (union required-features
 821                          '(=jis-x0208
 822                            =jis-x0208@1990
 823                            =jis-x0213-1@2000
 824                            =jis-x0213-1@2004
 825                            =jis-x0213-2
 826                            =jis-x0212
 827                            =jis-x0208@1983
 828                            =jis-x0208@1978
 829                            =shinjigen
 830                            =shinjigen@1ed
 831                            =shinjigen@rev
 832                            =shinjigen/+p@rev))))
 833            ((eq source 'CN)
 834             (setq required-features
 835                   (union required-features
 836                          '(=gb2312
 837                            =gb12345
 838                            =iso-ir165)))))
 839           (cond
 840            ((find-charset (setq ret (intern (format "=%s" source))))
 841             (setq required-features
 842                   (cons ret required-features)))
 843            (t (setq required-features
 844                     (cons source required-features)))))
 845         (cond ((string-match "@JP" (symbol-name name))
 846                (setq required-features
 847                      (union required-features
 848                             '(=jis-x0208
 849                               =jis-x0208@1990
 850                               =jis-x0213-1-2000
 851                               =jis-x0213-2-2000
 852                               =jis-x0212
 853                               =jis-x0208@1983
 854                               =jis-x0208@1978))))
 855               ((string-match "@CN" (symbol-name name))
 856                (setq required-features
 857                      (union required-features
 858                             '(=gb2312
 859                               =gb12345
 860                               =iso-ir165)))))
 861         (if separator
 862             (insert lbs))
 863         (if readable
 864             (insert (format "%S" cell))
 865           (char-db-insert-char-spec cell readable
 866                                     nil
 867                                     required-features))
 868         (setq separator lbs))
 869        ((consp cell)
 870         (if separator
 871             (insert lbs))
 872         (if (consp (car cell))
 873             (char-db-insert-char-spec cell readable)
 874           (char-db-insert-char-reference cell readable))
 875         (setq separator lbs))
 876        (t
 877         (if separator
 878             (insert separator))
 879         (insert (prin1-to-string cell))
 880         (setq separator " ")))
 881       (setq value (cdr value)))
 882     (insert ")")
 883     (insert line-breaking)))
 884
 885 (defun insert-char-attributes (char &optional readable attributes column
 886                                     for-sub-node)
 887   (unless column
 888     (setq column (current-column)))
 889   (let (name value ; has-long-ccs-name
 890         rest
 891         radical strokes
 892         (line-breaking
 893          (concat "\n" (make-string (1+ column) ?\ )))
 894         lbs cell separator ret
 895         key al cal
 896         dest-ccss ; sources required-features
 897         ccss)
 898     (let (atr-d)
 899       (setq attributes
 900             (sort (if attributes
 901                       (if (consp attributes)
 902                           (progn
 903                             (dolist (name attributes)
 904                               (unless (memq name char-db-ignored-attributes)
 905                                 (if (find-charset name)
 906                                     (push name ccss))
 907                                 (push name atr-d)))
 908                             atr-d))
 909                     (dolist (name (char-attribute-list))
 910                       (unless (memq name char-db-ignored-attributes)
 911                         (if (find-charset name)
 912                             (push name ccss))
 913                         (push name atr-d)))
 914                     atr-d)
 915                   #'char-attribute-name<)))
 916     (insert "(")
 917     (when (memq '<-subsumptive attributes)
 918       (when (or readable (not for-sub-node))
 919         (when (setq value (get-char-attribute char '<-subsumptive))
 920           (char-db-insert-relation-feature char '<-subsumptive value
 921                                            line-breaking
 922                                            ccss readable)))
 923       (setq attributes (delq '<-subsumptive attributes)))
 924     (when (and (memq '<-denotational attributes)
 925                (setq value (get-char-attribute char '<-denotational)))
 926       (char-db-insert-relation-feature char '<-denotational value
 927                                        line-breaking
 928                                        ccss readable)
 929       (setq attributes (delq '<-denotational attributes)))
 930     (when (and (memq '<-denotational@component attributes)
 931                (setq value (get-char-attribute char '<-denotational@component)))
 932       (char-db-insert-relation-feature char '<-denotational@component value
 933                                        line-breaking
 934                                        ccss readable)
 935       (setq attributes (delq '<-denotational@component attributes)))
 936     (when (and (memq 'name attributes)
 937                (setq value (get-char-attribute char 'name)))
 938       (insert (format
 939                (if (> (+ (current-column) (length value)) 48)
 940                    "(name . %S)%s"
 941                  "(name               . %S)%s")
 942                value line-breaking))
 943       (setq attributes (delq 'name attributes))
 944       )
 945     (when (and (memq 'name* attributes)
 946                (setq value (get-char-attribute char 'name*)))
 947       (insert (format
 948                (if (> (+ (current-column) (length value)) 48)
 949                    "(name* . %S)%s"
 950                  "(name*              . %S)%s")
 951                value line-breaking))
 952       (setq attributes (delq 'name* attributes))
 953       )
 954     (when (and (memq 'script attributes)
 955                (setq value (get-char-attribute char 'script)))
 956       (insert (format "(script\t\t%s)%s"
 957                       (mapconcat (function prin1-to-string)
 958                                  value " ")
 959                       line-breaking))
 960       (setq attributes (delq 'script attributes))
 961       )
 962     (dolist (name '(=>ucs =>ucs*))
 963       (when (and (memq name attributes)
 964                  (setq value (get-char-attribute char name)))
 965         (insert (format "(%-18s . #x%04X)\t; %c%s"
 966                         name value (decode-char '=ucs value)
 967                         line-breaking))
 968         (setq attributes (delq name attributes))))
 969     (dolist (name '(=>ucs@gb =>ucs@big5))
 970       (when (and (memq name attributes)
 971                  (setq value (get-char-attribute char name)))
 972         (insert (format "(%-18s . #x%04X)\t; %c%s"
 973                         name value
 974                         (decode-char (intern
 975                                       (concat "="
 976                                               (substring
 977                                                (symbol-name name) 2)))
 978                                      value)
 979                         line-breaking))
 980         (setq attributes (delq name attributes))
 981         ))
 982     ;; (dolist (name '(=>daikanwa))
 983     ;;   (when (and (memq name attributes)
 984     ;;              (setq value (get-char-attribute char name)))
 985     ;;     (insert
 986     ;;      (if (integerp value)
 987     ;;          (format "(%-18s . %05d)\t; %c%s"
 988     ;;                  name value (decode-char '=daikanwa value)
 989     ;;                  line-breaking)
 990     ;;        (format "(%-18s %s)\t; %c%s"
 991     ;;                name
 992     ;;                (mapconcat (function prin1-to-string)
 993     ;;                           value " ")
 994     ;;                (char-representative-of-daikanwa char)
 995     ;;                line-breaking)))
 996     ;;     (setq attributes (delq name attributes))))
 997     (when (and (memq 'general-category attributes)
 998                (setq value (get-char-attribute char 'general-category)))
 999       (insert (format
1000                "(general-category\t%s) ; %s%s"
1001                (mapconcat (lambda (cell)
1002                             (format "%S" cell))
1003                           value " ")
1004                (cond ((rassoc value unidata-normative-category-alist)
1005                       "Normative Category")
1006                      ((rassoc value unidata-informative-category-alist)
1007                       "Informative Category")
1008                      (t
1009                       "Unknown Category"))
1010                line-breaking))
1011       (setq attributes (delq 'general-category attributes))
1012       )
1013     (when (and (memq 'bidi-category attributes)
1014                (setq value (get-char-attribute char 'bidi-category)))
1015       (insert (format "(bidi-category\t. %S)%s"
1016                       value
1017                       line-breaking))
1018       (setq attributes (delq 'bidi-category attributes))
1019       )
1020     (unless (or (not (memq 'mirrored attributes))
1021                 (eq (setq value (get-char-attribute char 'mirrored 'empty))
1022                     'empty))
1023       (insert (format "(mirrored\t\t. %S)%s"
1024                       value
1025                       line-breaking))
1026       (setq attributes (delq 'mirrored attributes))
1027       )
1028     (cond
1029      ((and (memq 'decimal-digit-value attributes)
1030            (setq value (get-char-attribute char 'decimal-digit-value)))
1031       (insert (format "(decimal-digit-value . %S)%s"
1032                       value
1033                       line-breaking))
1034       (setq attributes (delq 'decimal-digit-value attributes))
1035       (when (and (memq 'digit-value attributes)
1036                  (setq value (get-char-attribute char 'digit-value)))
1037         (insert (format "(digit-value\t . %S)%s"
1038                         value
1039                         line-breaking))
1040         (setq attributes (delq 'digit-value attributes))
1041         )
1042       (when (and (memq 'numeric-value attributes)
1043                  (setq value (get-char-attribute char 'numeric-value)))
1044         (insert (format "(numeric-value\t . %S)%s"
1045                         value
1046                         line-breaking))
1047         (setq attributes (delq 'numeric-value attributes))
1048         )
1049       )
1050      (t
1051       (when (and (memq 'digit-value attributes)
1052                  (setq value (get-char-attribute char 'digit-value)))
1053         (insert (format "(digit-value\t. %S)%s"
1054                         value
1055                         line-breaking))
1056         (setq attributes (delq 'digit-value attributes))
1057         )
1058       (when (and (memq 'numeric-value attributes)
1059                  (setq value (get-char-attribute char 'numeric-value)))
1060         (insert (format "(numeric-value\t. %S)%s"
1061                         value
1062                         line-breaking))
1063         (setq attributes (delq 'numeric-value attributes))
1064         )))
1065     (when (and (memq 'iso-10646-comment attributes)
1066                (setq value (get-char-attribute char 'iso-10646-comment)))
1067       (insert (format "(iso-10646-comment\t. %S)%s"
1068                       value
1069                       line-breaking))
1070       (setq attributes (delq 'iso-10646-comment attributes))
1071       )
1072     (when (and (memq 'morohashi-daikanwa attributes)
1073                (setq value (get-char-attribute char 'morohashi-daikanwa)))
1074       (insert (format "(morohashi-daikanwa\t%s)%s"
1075                       (mapconcat (function prin1-to-string) value " ")
1076                       line-breaking))
1077       (setq attributes (delq 'morohashi-daikanwa attributes))
1078       )
1079     (setq radical nil
1080           strokes nil)
1081     (when (and (memq 'ideographic-radical attributes)
1082                (setq value (get-char-attribute char 'ideographic-radical)))
1083       (setq radical value)
1084       (insert (format "(ideographic-radical . %S)\t; %c%s"
1085                       radical
1086                       (ideographic-radical radical)
1087                       line-breaking))
1088       (setq attributes (delq 'ideographic-radical attributes))
1089       )
1090     (when (and (memq 'shuowen-radical attributes)
1091                (setq value (get-char-attribute char 'shuowen-radical)))
1092       (insert (format "(shuowen-radical\t. %S)\t; %c%s"
1093                       value
1094                       (shuowen-radical value)
1095                       line-breaking))
1096       (setq attributes (delq 'shuowen-radical attributes))
1097       )
1098     (let (key)
1099       (dolist (domain
1100                (append
1101                 char-db-feature-domains
1102                 (let (dest domain)
1103                   (dolist (feature (char-attribute-list))
1104                     (setq feature (symbol-name feature))
1105                     (when (string-match
1106                            "\\(radical\\|strokes\\)@\\([^@*]+\\)\\(\\*\\|$\\)"
1107                            feature)
1108                       (setq domain (intern (match-string 2 feature)))
1109                      (unless (memq domain dest)
1110                        (setq dest (cons domain dest)))))
1111                   (sort dest #'string<))))
1112         (setq key (intern (format "%s@%s" 'ideographic-radical domain)))
1113         (when (and (memq key attributes)
1114                    (setq value (get-char-attribute char key)))
1115           (setq radical value)
1116           (insert (format "(%s . %S)\t; %c%s"
1117                           key
1118                           radical
1119                           (ideographic-radical radical)
1120                           line-breaking))
1121           (setq attributes (delq key attributes))
1122           )
1123         (setq key (intern (format "%s@%s" 'ideographic-strokes domain)))
1124         (when (and (memq key attributes)
1125                    (setq value (get-char-attribute char key)))
1126           (setq strokes value)
1127           (insert (format "(%s . %S)%s"
1128                           key
1129                           strokes
1130                           line-breaking))
1131           (setq attributes (delq key attributes))
1132           )
1133         (setq key (intern (format "%s@%s" 'total-strokes domain)))
1134         (when (and (memq key attributes)
1135                    (setq value (get-char-attribute char key)))
1136           (insert (format "(%s       . %S)%s"
1137                           key
1138                           value
1139                           line-breaking))
1140           (setq attributes (delq key attributes))
1141           )
1142         (dolist (feature '(ideographic-radical
1143                            ideographic-strokes
1144                            total-strokes))
1145           (setq key (intern (format "%s@%s*sources" feature domain)))
1146           (when (and (memq key attributes)
1147                      (setq value (get-char-attribute char key)))
1148             (insert (format "(%s%s" key line-breaking))
1149             (dolist (cell value)
1150               (insert (format " %s" cell)))
1151             (insert ")")
1152             (insert line-breaking)
1153             (setq attributes (delq key attributes))
1154             ))
1155         ))
1156     (when (and (memq 'ideographic-strokes attributes)
1157                (setq value (get-char-attribute char 'ideographic-strokes)))
1158       (setq strokes value)
1159       (insert (format "(ideographic-strokes . %S)%s"
1160                       strokes
1161                       line-breaking))
1162       (setq attributes (delq 'ideographic-strokes attributes))
1163       )
1164     (when (and (memq 'kangxi-radical attributes)
1165                (setq value (get-char-attribute char 'kangxi-radical)))
1166       (unless (eq value radical)
1167         (insert (format "(kangxi-radical\t . %S)\t; %c%s"
1168                         value
1169                         (ideographic-radical value)
1170                         line-breaking))
1171         (or radical
1172             (setq radical value)))
1173       (setq attributes (delq 'kangxi-radical attributes))
1174       )
1175     (when (and (memq 'kangxi-strokes attributes)
1176                (setq value (get-char-attribute char 'kangxi-strokes)))
1177       (unless (eq value strokes)
1178         (insert (format "(kangxi-strokes\t . %S)%s"
1179                         value
1180                         line-breaking))
1181         (or strokes
1182             (setq strokes value)))
1183       (setq attributes (delq 'kangxi-strokes attributes))
1184       )
1185     ;; (when (and (memq 'japanese-radical attributes)
1186     ;;            (setq value (get-char-attribute char 'japanese-radical)))
1187     ;;   (unless (eq value radical)
1188     ;;     (insert (format "(japanese-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 'japanese-radical attributes))
1195     ;;   )
1196     (when (and (memq 'japanese-strokes attributes)
1197                (setq value (get-char-attribute char 'japanese-strokes)))
1198       (unless (eq value strokes)
1199         (insert (format "(japanese-strokes\t . %S)%s"
1200                         value
1201                         line-breaking))
1202         (or strokes
1203             (setq strokes value)))
1204       (setq attributes (delq 'japanese-strokes attributes))
1205       )
1206     (when (and (memq 'cns-radical attributes)
1207                (setq value (get-char-attribute char 'cns-radical)))
1208       (insert (format "(cns-radical\t . %S)\t; %c%s"
1209                       value
1210                       (ideographic-radical value)
1211                       line-breaking))
1212       (setq attributes (delq 'cns-radical attributes))
1213       )
1214     (when (and (memq 'cns-strokes attributes)
1215                (setq value (get-char-attribute char 'cns-strokes)))
1216       (unless (eq value strokes)
1217         (insert (format "(cns-strokes\t . %S)%s"
1218                         value
1219                         line-breaking))
1220         (or strokes
1221             (setq strokes value)))
1222       (setq attributes (delq 'cns-strokes attributes))
1223       )
1224     ;; (when (and (memq 'shinjigen-1-radical attributes)
1225     ;;            (setq value (get-char-attribute char 'shinjigen-1-radical)))
1226     ;;   (unless (eq value radical)
1227     ;;     (insert (format "(shinjigen-1-radical . %S)\t; %c%s"
1228     ;;                     value
1229     ;;                     (ideographic-radical value)
1230     ;;                     line-breaking))
1231     ;;     (or radical
1232     ;;         (setq radical value)))
1233     ;;   (setq attributes (delq 'shinjigen-1-radical attributes))
1234     ;;   )
1235     ;; (when (and (memq 'ideographic- attributes)
1236     ;;            (setq value (get-char-attribute char 'ideographic-)))
1237     ;;   (insert "(ideographic-       ")
1238     ;;   (setq lbs (concat "\n" (make-string (current-column) ?\ ))
1239     ;;         separator nil)
1240     ;;   (while (consp value)
1241     ;;     (setq cell (car value))
1242     ;;     (if (integerp cell)
1243     ;;         (setq cell (decode-char '=ucs cell)))
1244     ;;     (cond ((characterp cell)
1245     ;;            (if separator
1246     ;;                (insert lbs))
1247     ;;            (if readable
1248     ;;                (insert (format "%S" cell))
1249     ;;              (char-db-insert-char-spec cell readable))
1250     ;;            (setq separator lbs))
1251     ;;           ((consp cell)
1252     ;;            (if separator
1253     ;;                (insert lbs))
1254     ;;            (if (consp (car cell))
1255     ;;                (char-db-insert-char-spec cell readable)
1256     ;;              (char-db-insert-char-reference cell readable))
1257     ;;            (setq separator lbs))
1258     ;;           (t
1259     ;;            (if separator
1260     ;;                (insert separator))
1261     ;;            (insert (prin1-to-string cell))
1262     ;;            (setq separator " ")))
1263     ;;     (setq value (cdr value)))
1264     ;;   (insert ")")
1265     ;;   (insert line-breaking)
1266     ;;   (setq attributes (delq 'ideographic- attributes)))
1267     (when (and (memq 'total-strokes attributes)
1268                (setq value (get-char-attribute char 'total-strokes)))
1269       (insert (format "(total-strokes       . %S)%s"
1270                       value
1271                       line-breaking))
1272       (setq attributes (delq 'total-strokes attributes))
1273       )
1274     ;; (when (and (memq '->ideograph attributes)
1275     ;;            (setq value (get-char-attribute char '->ideograph)))
1276     ;;   (insert (format "(->ideograph\t%s)%s"
1277     ;;                   (mapconcat (lambda (code)
1278     ;;                                (cond ((symbolp code)
1279     ;;                                       (symbol-name code))
1280     ;;                                      ((integerp code)
1281     ;;                                       (format "#x%04X" code))
1282     ;;                                      (t
1283     ;;                                       (format "%s %S"
1284     ;;                                               line-breaking code))))
1285     ;;                              value " ")
1286     ;;                   line-breaking))
1287     ;;   (setq attributes (delq '->ideograph attributes))
1288     ;;   )
1289     ;; (when (and (memq '->decomposition attributes)
1290     ;;            (setq value (get-char-attribute char '->decomposition)))
1291     ;;   (insert (format "(->decomposition\t%s)%s"
1292     ;;                   (mapconcat (lambda (code)
1293     ;;                                (cond ((symbolp code)
1294     ;;                                       (symbol-name code))
1295     ;;                                      ((characterp code)
1296     ;;                                       (if readable
1297     ;;                                           (format "%S" code)
1298     ;;                                         (format "#x%04X"
1299     ;;                                                 (char-int code))
1300     ;;                                         ))
1301     ;;                                      ((integerp code)
1302     ;;                                       (format "#x%04X" code))
1303     ;;                                      (t
1304     ;;                                       (format "%s%S" line-breaking code))))
1305     ;;                              value " ")
1306     ;;                   line-breaking))
1307     ;;   (setq attributes (delq '->decomposition attributes))
1308     ;;   )
1309     (if (equal (get-char-attribute char '->titlecase)
1310                (get-char-attribute char '->uppercase))
1311         (setq attributes (delq '->titlecase attributes)))
1312     (when (and (memq '->mojikyo attributes)
1313                (setq value (get-char-attribute char '->mojikyo)))
1314       (insert (format "(->mojikyo\t\t. %06d)\t; %c%s"
1315                       value (decode-char 'mojikyo value)
1316                       line-breaking))
1317       (setq attributes (delq '->mojikyo attributes))
1318       )
1319     ;; (when (and (memq 'hanyu-dazidian-vol attributes)
1320     ;;            (setq value (get-char-attribute char 'hanyu-dazidian-vol)))
1321     ;;   (insert (format "(hanyu-dazidian-vol  . %d)%s"
1322     ;;                   value line-breaking))
1323     ;;   (setq attributes (delq 'hanyu-dazidian-vol attributes))
1324     ;;   )
1325     ;; (when (and (memq 'hanyu-dazidian-page attributes)
1326     ;;            (setq value (get-char-attribute char 'hanyu-dazidian-page)))
1327     ;;   (insert (format "(hanyu-dazidian-page . %d)%s"
1328     ;;                   value line-breaking))
1329     ;;   (setq attributes (delq 'hanyu-dazidian-page attributes))
1330     ;;   )
1331     ;; (when (and (memq 'hanyu-dazidian-char attributes)
1332     ;;            (setq value (get-char-attribute char 'hanyu-dazidian-char)))
1333     ;;   (insert (format "(hanyu-dazidian-char . %d)%s"
1334     ;;                   value line-breaking))
1335     ;;   (setq attributes (delq 'hanyu-dazidian-char attributes))
1336     ;;   )
1337     (unless readable
1338       (dolist (ignored '(composition
1339                          ->denotational <-subsumptive ->ucs-unified
1340                          ->ideographic-component-forms))
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