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