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-006
 337     =ucs-var-008
 338     =ucs-var-010
 339     =ucs-itaiji-001
 340     =ucs-itaiji-002
 341     =ucs-itaiji-003
 342     =ucs-itaiji-004
 343     =ucs-itaiji-005
 344     =ucs-itaiji-006
 345     =ucs-itaiji-008
 346     =ucs-itaiji-009
 347     =>ucs-itaiji-001
 348     =>ucs-itaiji-005
 349     =>big5-cdp-itaiji-001
 350     =big5-cdp-itaiji-001
 351     =big5-cdp-itaiji-002
 352     =big5-cdp-itaiji-003
 353     =big5-cdp-var-003
 354     =big5-cdp-var-004
 355     =big5-cdp-var-005
 356     =>ucs@iwds-1
 357     =>ucs@component
 358     ==ucs@gb
 359     =ucs@JP/hanazono))
 360
 361
 362 ;;; @ char-db formatters
 363 ;;;
 364
 365 (defun char-db-make-char-spec (char)
 366   (let (ret char-spec)
 367     (cond ((characterp char)
 368            (cond ((and (setq ret (encode-char char '=ucs 'defined-only))
 369                        (not (and (<= #xE000 ret)(<= ret #xF8FF))))
 370                   (setq char-spec (list (cons '=ucs ret)))
 371                   (cond ((setq ret (get-char-attribute char 'name))
 372                          (setq char-spec (cons (cons 'name ret) char-spec))
 373                          )
 374                         ((setq ret (get-char-attribute char 'name*))
 375                          (setq char-spec (cons (cons 'name* ret) char-spec))
 376                          ))
 377                   )
 378                  ((encode-char char '=mj 'defined-only)
 379                   (setq char-spec nil)
 380                   (dolist (ccs (charset-list))
 381                     (if (and (or (eq ccs '=mj)
 382                                  ;; (eq (charset-property ccs 'iso-ir) 177)
 383                                  (string-match "=ucs@" (symbol-name ccs))
 384                                  )
 385                              (setq ccs (charset-name ccs))
 386                              (null (assq ccs char-spec))
 387                              (setq ret (encode-char char ccs 'defined-only)))
 388                         (setq char-spec (cons (cons ccs ret) char-spec))))
 389                   )
 390                  ((encode-char char '==mj 'defined-only)
 391                   (setq char-spec nil)
 392                   (dolist (ccs (charset-list))
 393                     (if (and (or (eq ccs '==mj)
 394                                  ;; (eq (charset-property ccs 'iso-ir) 177)
 395                                  (string-match "=ucs@" (symbol-name ccs))
 396                                  )
 397                              (setq ccs (charset-name ccs))
 398                              (null (assq ccs char-spec))
 399                              (setq ret (encode-char char ccs 'defined-only)))
 400                         (setq char-spec (cons (cons ccs ret) char-spec))))
 401                   )
 402                  ((encode-char char '=adobe-japan1 'defined-only)
 403                   (setq char-spec nil)
 404                   (dolist (ccs (charset-list))
 405                     (if (and (or (memq ccs
 406                                        '(=adobe-japan1-0
 407                                          =adobe-japan1-1
 408                                          =adobe-japan1-2
 409                                          =adobe-japan1-3
 410                                          =adobe-japan1-4
 411                                          =adobe-japan1-5
 412                                          =adobe-japan1-6
 413                                          ))
 414                                  ;; (eq (charset-property ccs 'iso-ir) 177)
 415                                  (string-match "=ucs@" (symbol-name ccs))
 416                                  )
 417                              (setq ccs (charset-name ccs))
 418                              (null (assq ccs char-spec))
 419                              (setq ret (encode-char char ccs 'defined-only)))
 420                         (setq char-spec (cons (cons ccs ret) char-spec))))
 421                   )
 422                  ((encode-char char '==adobe-japan1 'defined-only)
 423                   (setq char-spec nil)
 424                   (dolist (ccs (charset-list))
 425                     (if (and (or (memq ccs
 426                                        '(==adobe-japan1-0
 427                                          ==adobe-japan1-1
 428                                          ==adobe-japan1-2
 429                                          ==adobe-japan1-3
 430                                          ==adobe-japan1-4
 431                                          ==adobe-japan1-5
 432                                          ==adobe-japan1-6
 433                                          ))
 434                                  ;; (eq (charset-property ccs 'iso-ir) 177)
 435                                  (string-match "=ucs@" (symbol-name ccs))
 436                                  )
 437                              (setq ccs (charset-name ccs))
 438                              (null (assq ccs char-spec))
 439                              (setq ret (encode-char char ccs 'defined-only)))
 440                         (setq char-spec (cons (cons ccs ret) char-spec))))
 441                   )
 442                  ((setq ret
 443                         (catch 'tag
 444                           (let ((rest char-db-coded-charset-priority-list)
 445                                 ccs)
 446                             (while rest
 447                               (setq ccs (charset-name
 448                                          (find-charset (car rest))))
 449                               (if (setq ret
 450                                         (encode-char char ccs
 451                                                      'defined-only))
 452                                   (throw 'tag (cons ccs ret)))
 453                               (setq rest (cdr rest))))))
 454                   (setq char-spec (list ret))
 455                   (dolist (ccs (delq (car ret) (charset-list)))
 456                     (if (and (or (charset-iso-final-char ccs)
 457                                  (memq ccs
 458                                        '(=daikanwa
 459                                          =daikanwa@rev2
 460                                          ;; =gt-k
 461                                          =jis-x0208@1997
 462                                          ))
 463                                  (eq (charset-property ccs 'iso-ir) 177)
 464                                  ;; (string-match "=ucs@" (symbol-name ccs))
 465                                  )
 466                              (setq ccs (charset-name ccs))
 467                              (null (assq ccs char-spec))
 468                              (setq ret (encode-char char ccs 'defined-only)))
 469                         (setq char-spec (cons (cons ccs ret) char-spec))))
 470                   (if (null char-spec)
 471                       (setq char-spec (split-char char)))
 472                   (cond ((setq ret (get-char-attribute char 'name))
 473                          (setq char-spec (cons (cons 'name ret) char-spec))
 474                          )
 475                         ((setq ret (get-char-attribute char 'name*))
 476                          (setq char-spec (cons (cons 'name* ret) char-spec))
 477                          ))
 478                   )
 479                  ((setq ret (get-char-attribute
 480                              char 'ideographic-combination))
 481                   (setq char-spec
 482                         (cons (cons 'ideographic-combination ret)
 483                               char-spec))
 484                   ))
 485            char-spec)
 486           ((consp char)
 487            char))))
 488
 489 (defun char-db-insert-char-spec (char &optional readable column
 490                                       required-features)
 491   (unless column
 492     (setq column (current-column)))
 493   (let (char-spec temp-char)
 494     (setq char-spec (char-db-make-char-spec char))
 495     (unless (or (characterp char) ; char
 496                 (condition-case nil
 497                     (setq char (find-char char-spec))
 498                   (error nil)))
 499       ;; define temporary character
 500       ;;   Current implementation is dirty.
 501       (setq temp-char (define-char (cons '(ideograph-daikanwa . 0)
 502                                          char-spec)))
 503       (remove-char-attribute temp-char 'ideograph-daikanwa)
 504       (setq char temp-char))
 505     (insert-char-attributes char
 506                             readable
 507                             (union (mapcar #'car char-spec)
 508                                    required-features)
 509                             nil 'for-sub-node)
 510     (when temp-char
 511       ;; undefine temporary character
 512       ;;   Current implementation is dirty.
 513       (setq char-spec (char-attribute-alist temp-char))
 514       (while char-spec
 515         (remove-char-attribute temp-char (car (car char-spec)))
 516         (setq char-spec (cdr char-spec))))))
 517
 518 (defun char-db-insert-alist (alist &optional readable column)
 519   (unless column
 520     (setq column (current-column)))
 521   (let ((line-breaking
 522          (concat "\n" (make-string (1+ column) ?\ )))
 523         name value
 524         ret al ; cal
 525         key
 526         lbs cell rest separator)
 527     (insert "(")
 528     (while alist
 529       (setq name (car (car alist))
 530             value (cdr (car alist)))
 531       (cond ((eq name 'char)
 532              (insert "(char . ")
 533              (if (setq ret (condition-case nil
 534                                (find-char value)
 535                              (error nil)))
 536                  (progn
 537                    (setq al nil
 538                          ;; cal nil
 539                          )
 540                    (while value
 541                      (setq key (car (car value)))
 542                      ;; (if (find-charset key)
 543                      ;;     (setq cal (cons key cal))
 544                      (setq al (cons key al))
 545                      ;; )
 546                      (setq value (cdr value)))
 547                    (insert-char-attributes ret
 548                                            readable
 549                                            (or al 'none) ; cal
 550                                            nil 'for-sub-node))
 551                (insert (prin1-to-string value)))
 552              (insert ")")
 553              (insert line-breaking))
 554             ((consp value)
 555              (insert (format "(%-18s " name))
 556              (setq lbs (concat "\n" (make-string (current-column) ?\ )))
 557              (while (consp value)
 558                (setq cell (car value))
 559                (if (and (consp cell)
 560                         (consp (car cell))
 561                         (setq ret (condition-case nil
 562                                       (find-char cell)
 563                                     (error nil)))
 564                         )
 565                    (progn
 566                      (setq rest cell
 567                            al nil
 568                            ;; cal nil
 569                            )
 570                      (while rest
 571                        (setq key (car (car rest)))
 572                        ;; (if (find-charset key)
 573                        ;;     (setq cal (cons key cal))
 574                        (setq al (cons key al))
 575                        ;; )
 576                        (setq rest (cdr rest)))
 577                      (if separator
 578                          (insert lbs))
 579                      (insert-char-attributes ret
 580                                              readable
 581                                              al ; cal
 582                                              nil 'for-sub-node)
 583                      (setq separator lbs))
 584                  (if separator
 585                      (insert separator))
 586                  (insert (prin1-to-string cell))
 587                  (setq separator " "))
 588                (setq value (cdr value)))
 589              (insert ")")
 590              (insert line-breaking))
 591             (t
 592              (insert (format "(%-18s . %S)%s"
 593                              name value
 594                              line-breaking))))
 595       (setq alist (cdr alist))))
 596   (insert ")"))
 597
 598 (defun char-db-insert-char-reference (plist &optional readable column)
 599   (unless column
 600     (setq column (current-column)))
 601   (let ((line-breaking
 602          (concat "\n" (make-string (1+ column) ?\ )))
 603         (separator "")
 604         name value)
 605     (insert "(")
 606     (while plist
 607       (setq name (pop plist))
 608       (setq value (pop plist))
 609       (cond ((eq name :char)
 610              (insert separator)
 611              (insert ":char\t")
 612              (cond ((numberp value)
 613                     (setq value (decode-char '=ucs value)))
 614                    ;; ((consp value)
 615                    ;;  (setq value (or (find-char value)
 616                    ;;                  value)))
 617                    )
 618              (char-db-insert-char-spec value readable)
 619              (insert line-breaking)
 620              (setq separator ""))
 621             ((eq name :radical)
 622              (insert (format "%s%s\t%d ; %c%s"
 623                              separator
 624                              name value
 625                              (ideographic-radical value)
 626                              line-breaking))
 627              (setq separator ""))
 628             (t
 629              (insert (format "%s%s\t%S" separator name value))
 630              (setq separator line-breaking)))
 631       ))
 632   (insert ")"))
 633
 634 (defun char-db-decode-isolated-char (ccs code-point)
 635   (let (ret)
 636     (setq ret
 637           (cond ((eq ccs 'arabic-iso8859-6)
 638                  (decode-char ccs code-point))
 639                 ;; ((eq ccs '=gt)
 640                 ;;  (decode-builtin-char '==gt code-point))
 641                 ((and (memq ccs '(=gt-pj-1
 642                                   =gt-pj-2
 643                                   =gt-pj-3
 644                                   =gt-pj-4
 645                                   =gt-pj-5
 646                                   =gt-pj-6
 647                                   =gt-pj-7
 648                                   =gt-pj-8
 649                                   =gt-pj-9
 650                                   =gt-pj-10
 651                                   =gt-pj-11))
 652                       (setq ret (decode-char ccs code-point))
 653                       (setq ret (encode-char ret '=gt 'defined-only)))
 654                  (decode-builtin-char '=gt ret))
 655                 (t
 656                  (decode-builtin-char ccs code-point))))
 657     (cond ((null ret)
 658            (or (decode-char ccs code-point)
 659                (define-char (list (cons ccs code-point)))))
 660           ((and (<= 0 (char-int ret))
 661                 (<= (char-int ret) #x1F))
 662            (decode-char '=ucs (+ #x2400 (char-int ret))))
 663           ((= (char-int ret) #x7F)
 664            ?\u2421)
 665           (t ret))))
 666
 667 (defvar char-db-convert-obsolete-format t)
 668
 669 (defun char-db-insert-ccs-feature (name value line-breaking)
 670   (cond
 671    ((integerp value)
 672     (insert
 673      (format
 674       (cond
 675        ((memq name '(=>iwds-1
 676                      ==shinjigen
 677                      =shinjigen
 678                      =shinjigen@1ed ==shinjigen@1ed
 679                      =shinjigen@rev ==shinjigen@rev
 680                      =shinjigen/+p@rev ==shinjigen/+p@rev
 681                      ===daikanwa/ho ==daikanwa/ho
 682                      =daikanwa/ho =>>daikanwa/ho =>daikanwa/ho))
 683         "(%-18s .  %04d)\t; %c")
 684        ((eq name '=shinjigen@1ed/24pr)
 685         "(%-18s . %04d)\t; %c")
 686        ((or
 687          (memq name
 688                '(===daikanwa
 689                  ==daikanwa =daikanwa =>>daikanwa =>daikanwa
 690                  =daikanwa@rev1 =daikanwa@rev2
 691                  =daikanwa/+p ==daikanwa/+p ===daikanwa/+p
 692                  =>>daikanwa/+p
 693                  =daikanwa/+2p =>>daikanwa/+2p
 694                  =gt ==gt ===gt
 695                  =>>gt =+>gt =>gt
 696                  =gt-k ==gt-k ===gt-k
 697                  =>>gt-k =>gt-k
 698                  =adobe-japan1-0 ==adobe-japan1-0 ===adobe-japan1-0
 699                  =adobe-japan1-1 ==adobe-japan1-1 ===adobe-japan1-1
 700                  =adobe-japan1-2 ==adobe-japan1-2 ===adobe-japan1-2
 701                  =adobe-japan1-3 ==adobe-japan1-3 ===adobe-japan1-3
 702                  =adobe-japan1-4 ==adobe-japan1-4 ===adobe-japan1-4
 703                  =adobe-japan1-5 ==adobe-japan1-5 ===adobe-japan1-5
 704                  =adobe-japan1-6 ==adobe-japan1-6 ===adobe-japan1-6
 705                  =>>adobe-japan1-0 =+>adobe-japan1-0
 706                  =>>adobe-japan1-1 =+>adobe-japan1-1
 707                  =>>adobe-japan1-2 =+>adobe-japan1-2
 708                  =>>adobe-japan1-3 =+>adobe-japan1-3
 709                  =>>adobe-japan1-4 =+>adobe-japan1-4
 710                  =>>adobe-japan1-5 =+>adobe-japan1-5
 711                  =>>adobe-japan1-6 =+>adobe-japan1-6
 712                  =>cbeta =cbeta =>>cbeta ==cbeta ===cbeta
 713                  =zinbun-oracle =>zinbun-oracle
 714                  ===hng-jou ===hng-keg ===hng-dng ===hng-mam
 715                  ===hng-drt ===hng-kgk ===hng-myz ===hng-kda
 716                  ===hng-khi ===hng-khm ===hng-hok ===hng-kyd ===hng-sok
 717                  ===hng-yhk ===hng-kak ===hng-kar ===hng-kae
 718                  ===hng-sys ===hng-tsu ===hng-tzj
 719                  ===hng-hos ===hng-nak ===hng-jhk
 720                  ===hng-hod ===hng-gok ===hng-ink ===hng-nto
 721                  ===hng-nkm ===hng-k24 ===hng-nkk
 722                  ===hng-kcc ===hng-kcj ===hng-kbk ===hng-sik
 723                  ===hng-skk ===hng-kyu ===hng-ksk ===hng-wan
 724                  ===hng-okd ===hng-wad ===hng-kmi
 725                  ===hng-zkd ===hng-doh ===hng-jyu
 726                  ===hng-tzs ===hng-kss ===hng-kyo
 727                  ===hng-smk))
 728          ;; (string-match "^=adobe-" (symbol-name name))
 729          )
 730         "(%-18s . %05d)\t; %c")
 731        ((memq name '(=hanyo-denshi/ks
 732                      ==hanyo-denshi/ks ===hanyo-denshi/ks
 733                      =>>hanyo-denshi/ks
 734                      =koseki ==koseki
 735                      =mj ==mj ===mj =>>mj =>mj
 736                      =zihai mojikyo))
 737         "(%-18s . %06d)\t; %c")
 738        ((memq name '(=hanyo-denshi/tk ==hanyo-denshi/tk))
 739         "(%-18s . %08d)\t; %c")
 740        ((>= (charset-dimension name) 2)
 741         "(%-18s . #x%04X)\t; %c")
 742        (t
 743         "(%-18s . #x%02X)\t; %c"))
 744       name
 745       (if (= (charset-iso-graphic-plane name) 1)
 746           (logior value
 747                   (cond ((= (charset-dimension name) 1)
 748                          #x80)
 749                         ((= (charset-dimension name) 2)
 750                          #x8080)
 751                         ((= (charset-dimension name) 3)
 752                          #x808080)
 753                         (t 0)))
 754         value)
 755       (char-db-decode-isolated-char name value)))
 756     (if (and (= (charset-chars name) 94)
 757              (= (charset-dimension name) 2))
 758         (insert (format " [%02d-%02d]"
 759                         (- (lsh value -8) 32)
 760                         (- (logand value 255) 32))))
 761     )
 762    (t
 763     (insert (format "(%-18s . %s)" name value))
 764     ))
 765   (insert line-breaking))
 766
 767 (defun char-db-insert-relation-feature (char name value line-breaking
 768                                              ccss readable)
 769   (insert (format "(%-18s%s " name line-breaking))
 770   (let ((lbs (concat "\n" (make-string (current-column) ?\ )))
 771         separator cell sources required-features
 772         ret)
 773     (while (consp value)
 774       (setq cell (car value))
 775       (if (integerp cell)
 776           (setq cell (decode-char '=ucs cell)))
 777       (cond
 778        ((eq name '->subsumptive)
 779         (when (or (not (some (lambda (atr)
 780                                (get-char-attribute cell atr))
 781                              char-db-ignored-attributes))
 782                   (some (lambda (ccs)
 783                           (encode-char cell ccs 'defined-only))
 784                         ccss))
 785           (if separator
 786               (insert lbs))
 787           (let ((char-db-ignored-attributes
 788                  (cons '<-subsumptive
 789                        char-db-ignored-attributes)))
 790             (insert-char-attributes cell readable nil nil 'for-sub-node))
 791           (setq separator lbs))
 792         )
 793        ((characterp cell)
 794         (setq sources
 795               (get-char-attribute
 796                char (intern (format "%s*sources" name))))
 797         (setq required-features nil)
 798         (dolist (source sources)
 799           (cond
 800            ((memq source '(JP
 801                            JP/Jouyou
 802                            shinjigen shinjigen@1ed shinjigen@rev))
 803             (setq required-features
 804                   (union required-features
 805                          '(=jis-x0208
 806                            =jis-x0208@1990
 807                            =jis-x0213-1@2000
 808                            =jis-x0213-1@2004
 809                            =jis-x0213-2
 810                            =jis-x0212
 811                            =jis-x0208@1983
 812                            =jis-x0208@1978
 813                            =shinjigen
 814                            =shinjigen@1ed
 815                            =shinjigen@rev
 816                            =shinjigen/+p@rev))))
 817            ((eq source 'CN)
 818             (setq required-features
 819                   (union required-features
 820                          '(=gb2312
 821                            =gb12345
 822                            =iso-ir165)))))
 823           (cond
 824            ((find-charset (setq ret (intern (format "=%s" source))))
 825             (setq required-features
 826                   (cons ret required-features)))
 827            (t (setq required-features
 828                     (cons source required-features)))))
 829         (cond ((string-match "@JP" (symbol-name name))
 830                (setq required-features
 831                      (union required-features
 832                             '(=jis-x0208
 833                               =jis-x0208@1990
 834                               =jis-x0213-1-2000
 835                               =jis-x0213-2-2000
 836                               =jis-x0212
 837                               =jis-x0208@1983
 838                               =jis-x0208@1978))))
 839               ((string-match "@CN" (symbol-name name))
 840                (setq required-features
 841                      (union required-features
 842                             '(=gb2312
 843                               =gb12345
 844                               =iso-ir165)))))
 845         (if separator
 846             (insert lbs))
 847         (if readable
 848             (insert (format "%S" cell))
 849           (char-db-insert-char-spec cell readable
 850                                     nil
 851                                     required-features))
 852         (setq separator lbs))
 853        ((consp cell)
 854         (if separator
 855             (insert lbs))
 856         (if (consp (car cell))
 857             (char-db-insert-char-spec cell readable)
 858           (char-db-insert-char-reference cell readable))
 859         (setq separator lbs))
 860        (t
 861         (if separator
 862             (insert separator))
 863         (insert (prin1-to-string cell))
 864         (setq separator " ")))
 865       (setq value (cdr value)))
 866     (insert ")")
 867     (insert line-breaking)))
 868
 869 (defun insert-char-attributes (char &optional readable attributes column
 870                                     for-sub-node)
 871   (unless column
 872     (setq column (current-column)))
 873   (let (name value ; has-long-ccs-name
 874         rest
 875         radical strokes
 876         (line-breaking
 877          (concat "\n" (make-string (1+ column) ?\ )))
 878         lbs cell separator ret
 879         key al cal
 880         dest-ccss ; sources required-features
 881         ccss)
 882     (let (atr-d)
 883       (setq attributes
 884             (sort (if attributes
 885                       (if (consp attributes)
 886                           (progn
 887                             (dolist (name attributes)
 888                               (unless (memq name char-db-ignored-attributes)
 889                                 (if (find-charset name)
 890                                     (push name ccss))
 891                                 (push name atr-d)))
 892                             atr-d))
 893                     (dolist (name (char-attribute-list))
 894                       (unless (memq name char-db-ignored-attributes)
 895                         (if (find-charset name)
 896                             (push name ccss))
 897                         (push name atr-d)))
 898                     atr-d)
 899                   #'char-attribute-name<)))
 900     (insert "(")
 901     (when (memq '<-subsumptive attributes)
 902       (when (or readable (not for-sub-node))
 903         (when (setq value (get-char-attribute char '<-subsumptive))
 904           (char-db-insert-relation-feature char '<-subsumptive value
 905                                            line-breaking
 906                                            ccss readable)))
 907       (setq attributes (delq '<-subsumptive attributes)))
 908     (when (and (memq '<-denotational attributes)
 909                (setq value (get-char-attribute char '<-denotational)))
 910       (char-db-insert-relation-feature char '<-denotational value
 911                                        line-breaking
 912                                        ccss readable)
 913       (setq attributes (delq '<-denotational attributes)))
 914     (when (and (memq '<-denotational@component attributes)
 915                (setq value (get-char-attribute char '<-denotational@component)))
 916       (char-db-insert-relation-feature char '<-denotational@component value
 917                                        line-breaking
 918                                        ccss readable)
 919       (setq attributes (delq '<-denotational@component attributes)))
 920     (when (and (memq 'name attributes)
 921                (setq value (get-char-attribute char 'name)))
 922       (insert (format
 923                (if (> (+ (current-column) (length value)) 48)
 924                    "(name . %S)%s"
 925                  "(name               . %S)%s")
 926                value line-breaking))
 927       (setq attributes (delq 'name attributes))
 928       )
 929     (when (and (memq 'name* attributes)
 930                (setq value (get-char-attribute char 'name*)))
 931       (insert (format
 932                (if (> (+ (current-column) (length value)) 48)
 933                    "(name* . %S)%s"
 934                  "(name*              . %S)%s")
 935                value line-breaking))
 936       (setq attributes (delq 'name* attributes))
 937       )
 938     (when (and (memq 'script attributes)
 939                (setq value (get-char-attribute char 'script)))
 940       (insert (format "(script\t\t%s)%s"
 941                       (mapconcat (function prin1-to-string)
 942                                  value " ")
 943                       line-breaking))
 944       (setq attributes (delq 'script attributes))
 945       )
 946     (dolist (name '(=>ucs =>ucs*))
 947       (when (and (memq name attributes)
 948                  (setq value (get-char-attribute char name)))
 949         (insert (format "(%-18s . #x%04X)\t; %c%s"
 950                         name value (decode-char '=ucs value)
 951                         line-breaking))
 952         (setq attributes (delq name attributes))))
 953     (dolist (name '(=>ucs@gb =>ucs@big5))
 954       (when (and (memq name attributes)
 955                  (setq value (get-char-attribute char name)))
 956         (insert (format "(%-18s . #x%04X)\t; %c%s"
 957                         name value
 958                         (decode-char (intern
 959                                       (concat "="
 960                                               (substring
 961                                                (symbol-name name) 2)))
 962                                      value)
 963                         line-breaking))
 964         (setq attributes (delq name attributes))
 965         ))
 966     ;; (dolist (name '(=>daikanwa))
 967     ;;   (when (and (memq name attributes)
 968     ;;              (setq value (get-char-attribute char name)))
 969     ;;     (insert
 970     ;;      (if (integerp value)
 971     ;;          (format "(%-18s . %05d)\t; %c%s"
 972     ;;                  name value (decode-char '=daikanwa value)
 973     ;;                  line-breaking)
 974     ;;        (format "(%-18s %s)\t; %c%s"
 975     ;;                name
 976     ;;                (mapconcat (function prin1-to-string)
 977     ;;                           value " ")
 978     ;;                (char-representative-of-daikanwa char)
 979     ;;                line-breaking)))
 980     ;;     (setq attributes (delq name attributes))))
 981     (when (and (memq 'general-category attributes)
 982                (setq value (get-char-attribute char 'general-category)))
 983       (insert (format
 984                "(general-category\t%s) ; %s%s"
 985                (mapconcat (lambda (cell)
 986                             (format "%S" cell))
 987                           value " ")
 988                (cond ((rassoc value unidata-normative-category-alist)
 989                       "Normative Category")
 990                      ((rassoc value unidata-informative-category-alist)
 991                       "Informative Category")
 992                      (t
 993                       "Unknown Category"))
 994                line-breaking))
 995       (setq attributes (delq 'general-category attributes))
 996       )
 997     (when (and (memq 'bidi-category attributes)
 998                (setq value (get-char-attribute char 'bidi-category)))
 999       (insert (format "(bidi-category\t. %S)%s"
1000                       value
1001                       line-breaking))
1002       (setq attributes (delq 'bidi-category attributes))
1003       )
1004     (unless (or (not (memq 'mirrored attributes))
1005                 (eq (setq value (get-char-attribute char 'mirrored 'empty))
1006                     'empty))
1007       (insert (format "(mirrored\t\t. %S)%s"
1008                       value
1009                       line-breaking))
1010       (setq attributes (delq 'mirrored attributes))
1011       )
1012     (cond
1013      ((and (memq 'decimal-digit-value attributes)
1014            (setq value (get-char-attribute char 'decimal-digit-value)))
1015       (insert (format "(decimal-digit-value . %S)%s"
1016                       value
1017                       line-breaking))
1018       (setq attributes (delq 'decimal-digit-value attributes))
1019       (when (and (memq 'digit-value attributes)
1020                  (setq value (get-char-attribute char 'digit-value)))
1021         (insert (format "(digit-value\t . %S)%s"
1022                         value
1023                         line-breaking))
1024         (setq attributes (delq 'digit-value attributes))
1025         )
1026       (when (and (memq 'numeric-value attributes)
1027                  (setq value (get-char-attribute char 'numeric-value)))
1028         (insert (format "(numeric-value\t . %S)%s"
1029                         value
1030                         line-breaking))
1031         (setq attributes (delq 'numeric-value attributes))
1032         )
1033       )
1034      (t
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     (when (and (memq 'iso-10646-comment attributes)
1050                (setq value (get-char-attribute char 'iso-10646-comment)))
1051       (insert (format "(iso-10646-comment\t. %S)%s"
1052                       value
1053                       line-breaking))
1054       (setq attributes (delq 'iso-10646-comment attributes))
1055       )
1056     (when (and (memq 'morohashi-daikanwa attributes)
1057                (setq value (get-char-attribute char 'morohashi-daikanwa)))
1058       (insert (format "(morohashi-daikanwa\t%s)%s"
1059                       (mapconcat (function prin1-to-string) value " ")
1060                       line-breaking))
1061       (setq attributes (delq 'morohashi-daikanwa attributes))
1062       )
1063     (setq radical nil
1064           strokes nil)
1065     (when (and (memq 'ideographic-radical attributes)
1066                (setq value (get-char-attribute char 'ideographic-radical)))
1067       (setq radical value)
1068       (insert (format "(ideographic-radical . %S)\t; %c%s"
1069                       radical
1070                       (ideographic-radical radical)
1071                       line-breaking))
1072       (setq attributes (delq 'ideographic-radical attributes))
1073       )
1074     (when (and (memq 'shuowen-radical attributes)
1075                (setq value (get-char-attribute char 'shuowen-radical)))
1076       (insert (format "(shuowen-radical\t. %S)\t; %c%s"
1077                       value
1078                       (shuowen-radical value)
1079                       line-breaking))
1080       (setq attributes (delq 'shuowen-radical attributes))
1081       )
1082     (let (key)
1083       (dolist (domain
1084                (append
1085                 char-db-feature-domains
1086                 (let (dest domain)
1087                   (dolist (feature (char-attribute-list))
1088                     (setq feature (symbol-name feature))
1089                     (when (string-match
1090                            "\\(radical\\|strokes\\)@\\([^@*]+\\)\\(\\*\\|$\\)"
1091                            feature)
1092                       (setq domain (intern (match-string 2 feature)))
1093                      (unless (memq domain dest)
1094                        (setq dest (cons domain dest)))))
1095                   (sort dest #'string<))))
1096         (setq key (intern (format "%s@%s" 'ideographic-radical domain)))
1097         (when (and (memq key attributes)
1098                    (setq value (get-char-attribute char key)))
1099           (setq radical value)
1100           (insert (format "(%s . %S)\t; %c%s"
1101                           key
1102                           radical
1103                           (ideographic-radical radical)
1104                           line-breaking))
1105           (setq attributes (delq key attributes))
1106           )
1107         (setq key (intern (format "%s@%s" 'ideographic-strokes domain)))
1108         (when (and (memq key attributes)
1109                    (setq value (get-char-attribute char key)))
1110           (setq strokes value)
1111           (insert (format "(%s . %S)%s"
1112                           key
1113                           strokes
1114                           line-breaking))
1115           (setq attributes (delq key attributes))
1116           )
1117         (setq key (intern (format "%s@%s" 'total-strokes domain)))
1118         (when (and (memq key attributes)
1119                    (setq value (get-char-attribute char key)))
1120           (insert (format "(%s       . %S)%s"
1121                           key
1122                           value
1123                           line-breaking))
1124           (setq attributes (delq key attributes))
1125           )
1126         (dolist (feature '(ideographic-radical
1127                            ideographic-strokes
1128                            total-strokes))
1129           (setq key (intern (format "%s@%s*sources" feature domain)))
1130           (when (and (memq key attributes)
1131                      (setq value (get-char-attribute char key)))
1132             (insert (format "(%s%s" key line-breaking))
1133             (dolist (cell value)
1134               (insert (format " %s" cell)))
1135             (insert ")")
1136             (insert line-breaking)
1137             (setq attributes (delq key attributes))
1138             ))
1139         ))
1140     (when (and (memq 'ideographic-strokes attributes)
1141                (setq value (get-char-attribute char 'ideographic-strokes)))
1142       (setq strokes value)
1143       (insert (format "(ideographic-strokes . %S)%s"
1144                       strokes
1145                       line-breaking))
1146       (setq attributes (delq 'ideographic-strokes attributes))
1147       )
1148     (when (and (memq 'kangxi-radical attributes)
1149                (setq value (get-char-attribute char 'kangxi-radical)))
1150       (unless (eq value radical)
1151         (insert (format "(kangxi-radical\t . %S)\t; %c%s"
1152                         value
1153                         (ideographic-radical value)
1154                         line-breaking))
1155         (or radical
1156             (setq radical value)))
1157       (setq attributes (delq 'kangxi-radical attributes))
1158       )
1159     (when (and (memq 'kangxi-strokes attributes)
1160                (setq value (get-char-attribute char 'kangxi-strokes)))
1161       (unless (eq value strokes)
1162         (insert (format "(kangxi-strokes\t . %S)%s"
1163                         value
1164                         line-breaking))
1165         (or strokes
1166             (setq strokes value)))
1167       (setq attributes (delq 'kangxi-strokes attributes))
1168       )
1169     ;; (when (and (memq 'japanese-radical attributes)
1170     ;;            (setq value (get-char-attribute char 'japanese-radical)))
1171     ;;   (unless (eq value radical)
1172     ;;     (insert (format "(japanese-radical\t . %S)\t; %c%s"
1173     ;;                     value
1174     ;;                     (ideographic-radical value)
1175     ;;                     line-breaking))
1176     ;;     (or radical
1177     ;;         (setq radical value)))
1178     ;;   (setq attributes (delq 'japanese-radical attributes))
1179     ;;   )
1180     (when (and (memq 'japanese-strokes attributes)
1181                (setq value (get-char-attribute char 'japanese-strokes)))
1182       (unless (eq value strokes)
1183         (insert (format "(japanese-strokes\t . %S)%s"
1184                         value
1185                         line-breaking))
1186         (or strokes
1187             (setq strokes value)))
1188       (setq attributes (delq 'japanese-strokes attributes))
1189       )
1190     (when (and (memq 'cns-radical attributes)
1191                (setq value (get-char-attribute char 'cns-radical)))
1192       (insert (format "(cns-radical\t . %S)\t; %c%s"
1193                       value
1194                       (ideographic-radical value)
1195                       line-breaking))
1196       (setq attributes (delq 'cns-radical attributes))
1197       )
1198     (when (and (memq 'cns-strokes attributes)
1199                (setq value (get-char-attribute char 'cns-strokes)))
1200       (unless (eq value strokes)
1201         (insert (format "(cns-strokes\t . %S)%s"
1202                         value
1203                         line-breaking))
1204         (or strokes
1205             (setq strokes value)))
1206       (setq attributes (delq 'cns-strokes attributes))
1207       )
1208     ;; (when (and (memq 'shinjigen-1-radical attributes)
1209     ;;            (setq value (get-char-attribute char 'shinjigen-1-radical)))
1210     ;;   (unless (eq value radical)
1211     ;;     (insert (format "(shinjigen-1-radical . %S)\t; %c%s"
1212     ;;                     value
1213     ;;                     (ideographic-radical value)
1214     ;;                     line-breaking))
1215     ;;     (or radical
1216     ;;         (setq radical value)))
1217     ;;   (setq attributes (delq 'shinjigen-1-radical attributes))
1218     ;;   )
1219     ;; (when (and (memq 'ideographic- attributes)
1220     ;;            (setq value (get-char-attribute char 'ideographic-)))
1221     ;;   (insert "(ideographic-       ")
1222     ;;   (setq lbs (concat "\n" (make-string (current-column) ?\ ))
1223     ;;         separator nil)
1224     ;;   (while (consp value)
1225     ;;     (setq cell (car value))
1226     ;;     (if (integerp cell)
1227     ;;         (setq cell (decode-char '=ucs cell)))
1228     ;;     (cond ((characterp cell)
1229     ;;            (if separator
1230     ;;                (insert lbs))
1231     ;;            (if readable
1232     ;;                (insert (format "%S" cell))
1233     ;;              (char-db-insert-char-spec cell readable))
1234     ;;            (setq separator lbs))
1235     ;;           ((consp cell)
1236     ;;            (if separator
1237     ;;                (insert lbs))
1238     ;;            (if (consp (car cell))
1239     ;;                (char-db-insert-char-spec cell readable)
1240     ;;              (char-db-insert-char-reference cell readable))
1241     ;;            (setq separator lbs))
1242     ;;           (t
1243     ;;            (if separator
1244     ;;                (insert separator))
1245     ;;            (insert (prin1-to-string cell))
1246     ;;            (setq separator " ")))
1247     ;;     (setq value (cdr value)))
1248     ;;   (insert ")")
1249     ;;   (insert line-breaking)
1250     ;;   (setq attributes (delq 'ideographic- attributes)))
1251     (when (and (memq 'total-strokes attributes)
1252                (setq value (get-char-attribute char 'total-strokes)))
1253       (insert (format "(total-strokes       . %S)%s"
1254                       value
1255                       line-breaking))
1256       (setq attributes (delq 'total-strokes attributes))
1257       )
1258     ;; (when (and (memq '->ideograph attributes)
1259     ;;            (setq value (get-char-attribute char '->ideograph)))
1260     ;;   (insert (format "(->ideograph\t%s)%s"
1261     ;;                   (mapconcat (lambda (code)
1262     ;;                                (cond ((symbolp code)
1263     ;;                                       (symbol-name code))
1264     ;;                                      ((integerp code)
1265     ;;                                       (format "#x%04X" code))
1266     ;;                                      (t
1267     ;;                                       (format "%s %S"
1268     ;;                                               line-breaking code))))
1269     ;;                              value " ")
1270     ;;                   line-breaking))
1271     ;;   (setq attributes (delq '->ideograph attributes))
1272     ;;   )
1273     ;; (when (and (memq '->decomposition attributes)
1274     ;;            (setq value (get-char-attribute char '->decomposition)))
1275     ;;   (insert (format "(->decomposition\t%s)%s"
1276     ;;                   (mapconcat (lambda (code)
1277     ;;                                (cond ((symbolp code)
1278     ;;                                       (symbol-name code))
1279     ;;                                      ((characterp code)
1280     ;;                                       (if readable
1281     ;;                                           (format "%S" code)
1282     ;;                                         (format "#x%04X"
1283     ;;                                                 (char-int code))
1284     ;;                                         ))
1285     ;;                                      ((integerp code)
1286     ;;                                       (format "#x%04X" code))
1287     ;;                                      (t
1288     ;;                                       (format "%s%S" line-breaking code))))
1289     ;;                              value " ")
1290     ;;                   line-breaking))
1291     ;;   (setq attributes (delq '->decomposition attributes))
1292     ;;   )
1293     (if (equal (get-char-attribute char '->titlecase)
1294                (get-char-attribute char '->uppercase))
1295         (setq attributes (delq '->titlecase attributes)))
1296     (when (and (memq '->mojikyo attributes)
1297                (setq value (get-char-attribute char '->mojikyo)))
1298       (insert (format "(->mojikyo\t\t. %06d)\t; %c%s"
1299                       value (decode-char 'mojikyo value)
1300                       line-breaking))
1301       (setq attributes (delq '->mojikyo attributes))
1302       )
1303     ;; (when (and (memq 'hanyu-dazidian-vol attributes)
1304     ;;            (setq value (get-char-attribute char 'hanyu-dazidian-vol)))
1305     ;;   (insert (format "(hanyu-dazidian-vol  . %d)%s"
1306     ;;                   value line-breaking))
1307     ;;   (setq attributes (delq 'hanyu-dazidian-vol attributes))
1308     ;;   )
1309     ;; (when (and (memq 'hanyu-dazidian-page attributes)
1310     ;;            (setq value (get-char-attribute char 'hanyu-dazidian-page)))
1311     ;;   (insert (format "(hanyu-dazidian-page . %d)%s"
1312     ;;                   value line-breaking))
1313     ;;   (setq attributes (delq 'hanyu-dazidian-page attributes))
1314     ;;   )
1315     ;; (when (and (memq 'hanyu-dazidian-char attributes)
1316     ;;            (setq value (get-char-attribute char 'hanyu-dazidian-char)))
1317     ;;   (insert (format "(hanyu-dazidian-char . %d)%s"
1318     ;;                   value line-breaking))
1319     ;;   (setq attributes (delq 'hanyu-dazidian-char attributes))
1320     ;;   )
1321     (unless readable
1322       (dolist (ignored '(composition
1323                          ->denotational <-subsumptive ->ucs-unified
1324                          ->ideographic-component-forms))
1325         (setq attributes (delq ignored attributes))))
1326     (while attributes
1327       (setq name (car attributes))
1328       (unless (eq (setq value (get-char-attribute char name 'value-is-empty))
1329                   'value-is-empty)
1330         (cond ((setq ret (find-charset name))
1331                (setq name (charset-name ret))
1332                (when (not (memq name dest-ccss))
1333                  (setq dest-ccss (cons name dest-ccss))
1334                  (char-db-insert-ccs-feature name value line-breaking))
1335                )
1336               ((string-match "^=>ucs@" (symbol-name name))
1337                (insert (format "(%-18s . #x%04X)\t; %c%s"
1338                                name value (decode-char '=ucs value)
1339                                line-breaking))
1340                )
1341               ((eq name 'jisx0208-1978/4X)
1342                (insert (format "(%-18s . #x%04X)%s"
1343                                name value
1344                                line-breaking))
1345                )
1346               ((and
1347                 (not readable)
1348                 (not (eq name '->subsumptive))
1349                 (not (eq name '->uppercase))
1350                 (not (eq name '->lowercase))
1351                 (not (eq name '->titlecase))
1352                 (not (eq name '->canonical))
1353                 (not (eq name '->Bopomofo))
1354                 (not (eq name '->mistakable))
1355                 (not (eq name '->ideographic-variants))
1356                 (null (get-char-attribute
1357                        char (intern (format "%s*sources" name))))
1358                 (not (string-match "\\*sources$" (symbol-name name)))
1359                 (null (get-char-attribute
1360                        char (intern (format "%s*note" name))))
1361                 (not (string-match "\\*note$" (symbol-name name)))
1362                 (or (eq name '<-identical)
1363                     (eq name '<-uppercase)
1364                     (eq name '<-lowercase)
1365                     (eq name '<-titlecase)
1366                     (eq name '<-canonical)
1367                     (eq name '<-ideographic-variants)
1368                     ;; (eq name '<-synonyms)
1369                     (string-match "^<-synonyms" (symbol-name name))
1370                     (eq name '<-mistakable)
1371                     (when (string-match "^->" (symbol-name name))
1372                       (cond
1373                        ((string-match "^->fullwidth" (symbol-name name))
1374                         (not (and (consp value)
1375                                   (characterp (car value))
1376                                   (encode-char
1377                                    (car value) '=ucs 'defined-only)))
1378                         )
1379                        (t)))
1380                     ))
1381                )
1382               ((or (eq name 'ideographic-structure)
1383                    (eq name 'ideographic-combination)
1384                    ;; (eq name 'ideographic-)
1385                    (eq name '=decomposition)
1386                    (char-feature-base-name= '=decomposition name)
1387                    (char-feature-base-name= '=>decomposition name)
1388                    ;; (string-match "^=>*decomposition\\(@[^*]+\\)?$"
1389                    ;;               (symbol-name name))
1390                    (string-match "^\\(->\\|<-\\)[^*]*$" (symbol-name name))
1391                    (string-match "^\\(->\\|<-\\)[^*]*\\*sources$"
1392                                  (symbol-name name))
1393                    )
1394                (char-db-insert-relation-feature char name value
1395                                                 line-breaking
1396                                                 ccss readable))
1397               ;; ((memq name '(ideograph=
1398               ;;               original-ideograph-of
1399               ;;               ancient-ideograph-of
1400               ;;               vulgar-ideograph-of
1401               ;;               wrong-ideograph-of
1402               ;;               ;; simplified-ideograph-of
1403               ;;               ideographic-variants
1404               ;;               ;; ideographic-different-form-of
1405               ;;               ))
1406               ;;  (insert (format "(%-18s%s " name line-breaking))
1407               ;;  (setq lbs (concat "\n" (make-string (current-column) ?\ ))
1408               ;;        separator nil)
1409               ;;  (while (consp value)
1410               ;;    (setq cell (car value))
1411               ;;    (if (and (consp cell)
1412               ;;             (consp (car cell)))
1413               ;;        (progn
1414               ;;          (if separator
1415               ;;              (insert lbs))
1416               ;;          (char-db-insert-alist cell readable)
1417               ;;          (setq separator lbs))
1418               ;;      (if separator
1419               ;;          (insert separator))
1420               ;;      (insert (prin1-to-string cell))
1421               ;;      (setq separator " "))
1422               ;;    (setq value (cdr value)))
1423               ;;  (insert ")")
1424               ;;  (insert line-breaking))
1425               ((consp value)
1426                (insert (format "(%-18s " name))
1427                (setq lbs (concat "\n" (make-string (current-column) ?\ ))
1428                      separator nil)
1429                (while (consp value)
1430                  (setq cell (car value))
1431                  (if (and (consp cell)
1432                           (consp (car cell))
1433                           (setq ret (condition-case nil
1434                                         (find-char cell)
1435                                       (error nil))))
1436                      (progn
1437                        (setq rest cell
1438                              al nil
1439                              cal nil)
1440                        (while rest
1441                          (setq key (car (car rest)))
1442                          (if (find-charset key)
1443                              (setq cal (cons key cal))
1444                            (setq al (cons key al)))
1445                          (setq rest (cdr rest)))
1446                        (if separator
1447                            (insert lbs))
1448                        (insert-char-attributes ret
1449                                                readable
1450                                                al ; cal
1451                                                nil 'for-sub-node)
1452                        (setq separator lbs))
1453                    (setq ret (prin1-to-string cell))
1454                    (if separator
1455                        (if (< (+ (current-column)
1456                                  (length ret)
1457                                  (length separator))
1458                               76)
1459                            (insert separator)
1460                          (insert lbs)))
1461                    (insert ret)
1462                    (setq separator " "))
1463                  (setq value (cdr value)))
1464                (insert ")")
1465                (insert line-breaking))
1466               (t
1467                (insert (format "(%-18s" name))
1468                (setq ret (prin1-to-string value))
1469                (unless (< (+ (current-column)
1470                              (length ret)
1471                              3)
1472                           76)
1473                  (insert line-breaking))
1474                (insert " . " ret ")" line-breaking)
1475                ;; (insert (format "(%-18s . %S)%s"
1476                ;;                 name value
1477                ;;                 line-breaking))
1478                )
1479               ))
1480       (setq attributes (cdr attributes)))
1481     (insert ")")))
1482
1483 (defun insert-char-data (char &optional readable
1484                               attributes)
1485   (save-restriction
1486     (narrow-to-region (point)(point))
1487     (insert "(define-char
1488   '")
1489     (insert-char-attributes char readable attributes)
1490     (insert ")\n")
1491     (goto-char (point-min))
1492     (while (re-search-forward "[ \t]+$" nil t)
1493       (replace-match ""))
1494     ;; from tabify.
1495     (goto-char (point-min))
1496     (while (re-search-forward "[ \t][ \t][ \t]*" nil t)
1497       (let ((column (current-column))
1498             (indent-tabs-mode t))
1499         (delete-region (match-beginning 0) (point))
1500         (indent-to column)))
1501     (goto-char (point-max))
1502     ;; (tabify (point-min)(point-max))
1503     ))
1504
1505 (defun insert-char-data-with-variant (char &optional printable
1506                                            no-ucs-unified
1507                                            script excluded-script)
1508   (insert-char-data char printable)
1509   (let ((variants (char-variants char))
1510         rest
1511         variant vs ret)
1512     (setq variants (sort variants #'<))
1513     (setq rest variants)
1514     (setq variants (cons char variants))
1515     (while rest
1516       (setq variant (car rest))
1517       (unless (get-char-attribute variant '<-subsumptive)
1518         (if (and (or (null script)
1519                      (null (setq vs (get-char-attribute variant 'script)))
1520                      (memq script vs))
1521                  (or (null excluded-script)
1522                      (null (setq vs (get-char-attribute variant 'script)))
1523                      (not (memq excluded-script vs))))
1524             (unless (and no-ucs-unified (get-char-attribute variant '=ucs))
1525               (insert-char-data variant printable)
1526               (if (setq ret (char-variants variant))
1527                   (while ret
1528                     (or (memq (car ret) variants)
1529                         ;; (get-char-attribute (car ret) '<-subsumptive)
1530                         (setq rest (nconc rest (list (car ret)))))
1531                     (setq ret (cdr ret)))))))
1532       (setq rest (cdr rest)))))
1533
1534 (defun insert-char-range-data (min max &optional script excluded-script)
1535   (let ((code min)
1536         char)
1537     (while (<= code max)
1538       (setq char (decode-char '=ucs code))
1539       (if (encode-char char '=ucs 'defined-only)
1540           (insert-char-data-with-variant char nil 'no-ucs-unified
1541                                          script excluded-script))
1542       (setq code (1+ code)))))
1543
1544 (defun write-char-range-data-to-file (min max file
1545                                           &optional script excluded-script)
1546   (let ((coding-system-for-write char-db-file-coding-system))
1547     (with-temp-buffer
1548       (insert (format ";; -*- coding: %s -*-\n"
1549                       char-db-file-coding-system))
1550       (insert-char-range-data min max script excluded-script)
1551       (write-region (point-min)(point-max) file))))
1552
1553 (defvar what-character-original-window-configuration)
1554
1555 ;;;###autoload
1556 (defun what-char-definition (char)
1557   (interactive (list (char-after)))
1558   (let ((buf (get-buffer-create "*Character Description*"))
1559         (the-buf (current-buffer))
1560         (win-conf (current-window-configuration)))
1561     (pop-to-buffer buf)
1562     (make-local-variable 'what-character-original-window-configuration)
1563     (setq what-character-original-window-configuration win-conf)
1564     (setq buffer-read-only nil)
1565     (erase-buffer)
1566     (condition-case err
1567         (progn
1568           (insert-char-data-with-variant char 'printable)
1569           (unless (char-attribute-alist char)
1570             (insert (format ";; = %c\n"
1571                             (let* ((rest (split-char char))
1572                                    (ccs (pop rest))
1573                                    (code (pop rest)))
1574                               (while rest
1575                                 (setq code (logior (lsh code 8)
1576                                                    (pop rest))))
1577                               (decode-char ccs code)))))
1578           ;; (char-db-update-comment)
1579           (set-buffer-modified-p nil)
1580           (view-mode the-buf (lambda (buf)
1581                                (set-window-configuration
1582                                 what-character-original-window-configuration)
1583                                ))
1584           (goto-char (point-min)))
1585       (error (progn
1586                (set-window-configuration
1587                 what-character-original-window-configuration)
1588                (signal (car err) (cdr err)))))))
1589
1590
1591 ;;; @ end
1592 ;;;
1593
1594 (provide 'char-db-util)
1595
1596 ;;; char-db-util.el ends here