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 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     ;; ==mj
 172     ;; ==adobe-japan1-0
 173     ;; ==adobe-japan1-1
 174     ;; ==adobe-japan1-2
 175     ;; ==adobe-japan1-3
 176     ;; ==adobe-japan1-4
 177     ;; ==adobe-japan1-5
 178     ;; ==adobe-japan1-6
 179     ==jis-x0208
 180     ==jis-x0213-1
 181     ==jis-x0213-2
 182     ==hanyo-denshi/ja
 183     ==hanyo-denshi/jb
 184     ==hanyo-denshi/jc
 185     ==hanyo-denshi/ft
 186     ==hanyo-denshi/ib
 187     ==hanyo-denshi/hg
 188     ==hanyo-denshi/jt
 189     ==hanyo-denshi/ks
 190     =gt
 191     =gt-k
 192     =daikanwa
 193     =daikanwa@rev2
 194     =daikanwa@rev1
 195     =daikanwa/+p
 196     ==daikanwa
 197     =cbeta
 198     ideograph-hanziku-1
 199     ideograph-hanziku-2
 200     ideograph-hanziku-3
 201     ideograph-hanziku-4
 202     ideograph-hanziku-5
 203     ideograph-hanziku-6
 204     ideograph-hanziku-7
 205     ideograph-hanziku-8
 206     ideograph-hanziku-9
 207     ideograph-hanziku-10
 208     ideograph-hanziku-11
 209     ideograph-hanziku-12
 210     ;; =>>>adobe-japan1-0
 211     ;; =>>>adobe-japan1-1
 212     ;; =>>>adobe-japan1-2
 213     ;; =>>>adobe-japan1-3
 214     ;; =>>>adobe-japan1-4
 215     ;; =>>>adobe-japan1-5
 216     ;; =>>>adobe-japan1-6
 217     ;; =>>>jis-x0208
 218     ;; =>>>jis-x0213-1
 219     ;; =>>>jis-x0213-2
 220     ;; =>>>hanyo-denshi/ja
 221     ;; =>>>hanyo-denshi/jb
 222     ;; =>>>hanyo-denshi/jc
 223     ;; =>>>hanyo-denshi/ft
 224     ;; =>>>hanyo-denshi/ib
 225     ;; =>>>hanyo-denshi/hg
 226     ;; =>>>hanyo-denshi/jt
 227     ;; =>>>hanyo-denshi/ks
 228     ;; =>>>gt
 229     =>>adobe-japan1-0
 230     =>>adobe-japan1-1
 231     =>>adobe-japan1-2
 232     =>>adobe-japan1-3
 233     =>>adobe-japan1-4
 234     =>>adobe-japan1-5
 235     =>>adobe-japan1-6
 236     =>>jis-x0208
 237     =>>jis-x0213-1
 238     =>>jis-x0213-1@2000
 239     =>>jis-x0213-1@2004
 240     =>>jis-x0213-2
 241     =>>jis-x0208@1978
 242     =>>hanyo-denshi/ft
 243     =>>hanyo-denshi/jt
 244     =>>hanyo-denshi/ks
 245     =>>gt
 246     =>>daikanwa
 247     =>>cbeta
 248     =+>jis-x0208
 249     =+>jis-x0213-1
 250     =+>jis-x0213-2
 251     =+>adobe-japan1-0
 252     =+>adobe-japan1-1
 253     =+>adobe-japan1-2
 254     =+>adobe-japan1-3
 255     =+>adobe-japan1-4
 256     =+>adobe-japan1-5
 257     =+>adobe-japan1-6
 258     =+>jis-x0208@1978
 259     =>jis-x0208
 260     =>jis-x0208@1997
 261     =>jis-x0213-1
 262     =>jis-x0213-1@2000
 263     =>jis-x0213-1@2004
 264     =>jis-x0213-2
 265     ==>ucs@bucs
 266     =>iwds-1
 267     =>ucs@hanyo-denshi
 268     =>ucs@iso
 269     =>ucs@unicode
 270     =>ucs@jis
 271     =>ucs@cns
 272     =>ucs@ks
 273     =+>ucs@iso
 274     =+>ucs@unicode
 275     =+>ucs@jis
 276     =+>ucs@jis/1990
 277     =+>ucs@cns
 278     =+>ucs@ks
 279     =>>ucs@iso
 280     =>>ucs@unicode
 281     =>>ucs@jis
 282     =>>ucs@cns
 283     =>>>ucs@iso
 284     =>>>ucs@unicode
 285     ==ucs@iso
 286     ==ucs@unicode
 287     ;; ==ucs@cns
 288     ==gb2312
 289     ==ks-x1001
 290     ==cns11643-1
 291     ==cns11643-2
 292     ==cns11643-3
 293     ==cns11643-4
 294     ==cns11643-5
 295     ==cns11643-6
 296     ==cns11643-7
 297     ==gt
 298     ==jis-x0208@1990
 299     ;; ==jis-x0208@1983
 300     ==jis-x0208@1978
 301     ==gt-k
 302     =ucs@iso
 303     =ucs@unicode
 304     =ucs@cns
 305     ==big5-cdp
 306     ==cbeta
 307     =>>big5-cdp
 308     =>>gt-k
 309     =+>gt
 310     =>gt
 311     =>big5-cdp
 312     =>daikanwa
 313     =>daikanwa/ho
 314     =>cns11643-5
 315     =>cns11643-7
 316     =big5
 317     =big5-eten
 318     =>gt-k
 319     =zinbun-oracle
 320     =>zinbun-oracle
 321     =ruimoku-v6
 322     =>>ruimoku-v6
 323     ==ruimoku-v6
 324     =jef-china3
 325     =>cbeta
 326     =shinjigen
 327     =ucs-var-001
 328     =ucs-var-002
 329     =ucs-var-003
 330     =ucs-itaiji-001
 331     =ucs-itaiji-002
 332     =ucs-itaiji-003
 333     =ucs-itaiji-004
 334     =ucs-itaiji-005
 335     =ucs-itaiji-008
 336     =big5-cdp-var-3
 337     =big5-cdp-var-5
 338     =>ucs@iwds-1))
 339
 340
 341 ;;; @ char-db formatters
 342 ;;;
 343
 344 (defun char-db-make-char-spec (char)
 345   (let (ret char-spec)
 346     (cond ((characterp char)
 347            (cond ((and (setq ret (encode-char char '=ucs 'defined-only))
 348                        (not (and (<= #xE000 ret)(<= ret #xF8FF))))
 349                   (setq char-spec (list (cons '=ucs ret)))
 350                   (cond ((setq ret (get-char-attribute char 'name))
 351                          (setq char-spec (cons (cons 'name ret) char-spec))
 352                          )
 353                         ((setq ret (get-char-attribute char 'name*))
 354                          (setq char-spec (cons (cons 'name* ret) char-spec))
 355                          ))
 356                   )
 357                  ((encode-char char '=mj 'defined-only)
 358                   (setq char-spec nil)
 359                   (dolist (ccs (charset-list))
 360                     (if (and (or (eq ccs '=mj)
 361                                  ;; (eq (charset-property ccs 'iso-ir) 177)
 362                                  (string-match "=ucs@" (symbol-name ccs))
 363                                  )
 364                              (setq ccs (charset-name ccs))
 365                              (null (assq ccs char-spec))
 366                              (setq ret (encode-char char ccs 'defined-only)))
 367                         (setq char-spec (cons (cons ccs ret) char-spec))))
 368                   )
 369                  ((encode-char char '==mj 'defined-only)
 370                   (setq char-spec nil)
 371                   (dolist (ccs (charset-list))
 372                     (if (and (or (eq ccs '==mj)
 373                                  ;; (eq (charset-property ccs 'iso-ir) 177)
 374                                  (string-match "=ucs@" (symbol-name ccs))
 375                                  )
 376                              (setq ccs (charset-name ccs))
 377                              (null (assq ccs char-spec))
 378                              (setq ret (encode-char char ccs 'defined-only)))
 379                         (setq char-spec (cons (cons ccs ret) char-spec))))
 380                   )
 381                  ((encode-char char '=adobe-japan1 'defined-only)
 382                   (setq char-spec nil)
 383                   (dolist (ccs (charset-list))
 384                     (if (and (or (memq ccs
 385                                        '(=adobe-japan1-0
 386                                          =adobe-japan1-1
 387                                          =adobe-japan1-2
 388                                          =adobe-japan1-3
 389                                          =adobe-japan1-4
 390                                          =adobe-japan1-5
 391                                          =adobe-japan1-6
 392                                          ))
 393                                  ;; (eq (charset-property ccs 'iso-ir) 177)
 394                                  (string-match "=ucs@" (symbol-name ccs))
 395                                  )
 396                              (setq ccs (charset-name ccs))
 397                              (null (assq ccs char-spec))
 398                              (setq ret (encode-char char ccs 'defined-only)))
 399                         (setq char-spec (cons (cons ccs ret) char-spec))))
 400                   )
 401                  ((encode-char char '==adobe-japan1 'defined-only)
 402                   (setq char-spec nil)
 403                   (dolist (ccs (charset-list))
 404                     (if (and (or (memq ccs
 405                                        '(==adobe-japan1-0
 406                                          ==adobe-japan1-1
 407                                          ==adobe-japan1-2
 408                                          ==adobe-japan1-3
 409                                          ==adobe-japan1-4
 410                                          ==adobe-japan1-5
 411                                          ==adobe-japan1-6
 412                                          ))
 413                                  ;; (eq (charset-property ccs 'iso-ir) 177)
 414                                  (string-match "=ucs@" (symbol-name ccs))
 415                                  )
 416                              (setq ccs (charset-name ccs))
 417                              (null (assq ccs char-spec))
 418                              (setq ret (encode-char char ccs 'defined-only)))
 419                         (setq char-spec (cons (cons ccs ret) char-spec))))
 420                   )
 421                  ((setq ret
 422                         (catch 'tag
 423                           (let ((rest char-db-coded-charset-priority-list)
 424                                 ccs)
 425                             (while rest
 426                               (setq ccs (charset-name
 427                                          (find-charset (car rest))))
 428                               (if (setq ret
 429                                         (encode-char char ccs
 430                                                      'defined-only))
 431                                   (throw 'tag (cons ccs ret)))
 432                               (setq rest (cdr rest))))))
 433                   (setq char-spec (list ret))
 434                   (dolist (ccs (delq (car ret) (charset-list)))
 435                     (if (and (or (charset-iso-final-char ccs)
 436                                  (memq ccs
 437                                        '(=daikanwa
 438                                          =daikanwa@rev2
 439                                          ;; =gt-k
 440                                          =jis-x0208@1997
 441                                          ))
 442                                  (eq (charset-property ccs 'iso-ir) 177)
 443                                  ;; (string-match "=ucs@" (symbol-name ccs))
 444                                  )
 445                              (setq ccs (charset-name ccs))
 446                              (null (assq ccs char-spec))
 447                              (setq ret (encode-char char ccs 'defined-only)))
 448                         (setq char-spec (cons (cons ccs ret) char-spec))))
 449                   (if (null char-spec)
 450                       (setq char-spec (split-char char)))
 451                   (cond ((setq ret (get-char-attribute char 'name))
 452                          (setq char-spec (cons (cons 'name ret) char-spec))
 453                          )
 454                         ((setq ret (get-char-attribute char 'name*))
 455                          (setq char-spec (cons (cons 'name* ret) char-spec))
 456                          ))
 457                   )
 458                  ((setq ret (get-char-attribute
 459                              char 'ideographic-combination))
 460                   (setq char-spec
 461                         (cons (cons 'ideographic-combination ret)
 462                               char-spec))
 463                   ))
 464            char-spec)
 465           ((consp char)
 466            char))))
 467
 468 (defun char-db-insert-char-spec (char &optional readable column
 469                                       required-features)
 470   (unless column
 471     (setq column (current-column)))
 472   (let (char-spec temp-char)
 473     (setq char-spec (char-db-make-char-spec char))
 474     (unless (or (characterp char) ; char
 475                 (condition-case nil
 476                     (setq char (find-char char-spec))
 477                   (error nil)))
 478       ;; define temporary character
 479       ;;   Current implementation is dirty.
 480       (setq temp-char (define-char (cons '(ideograph-daikanwa . 0)
 481                                          char-spec)))
 482       (remove-char-attribute temp-char 'ideograph-daikanwa)
 483       (setq char temp-char))
 484     (insert-char-attributes char
 485                             readable
 486                             (union (mapcar #'car char-spec)
 487                                    required-features)
 488                             nil 'for-sub-node)
 489     (when temp-char
 490       ;; undefine temporary character
 491       ;;   Current implementation is dirty.
 492       (setq char-spec (char-attribute-alist temp-char))
 493       (while char-spec
 494         (remove-char-attribute temp-char (car (car char-spec)))
 495         (setq char-spec (cdr char-spec))))))
 496
 497 (defun char-db-insert-alist (alist &optional readable column)
 498   (unless column
 499     (setq column (current-column)))
 500   (let ((line-breaking
 501          (concat "\n" (make-string (1+ column) ?\ )))
 502         name value
 503         ret al ; cal
 504         key
 505         lbs cell rest separator)
 506     (insert "(")
 507     (while alist
 508       (setq name (car (car alist))
 509             value (cdr (car alist)))
 510       (cond ((eq name 'char)
 511              (insert "(char . ")
 512              (if (setq ret (condition-case nil
 513                                (find-char value)
 514                              (error nil)))
 515                  (progn
 516                    (setq al nil
 517                          ;; cal nil
 518                          )
 519                    (while value
 520                      (setq key (car (car value)))
 521                      ;; (if (find-charset key)
 522                      ;;     (setq cal (cons key cal))
 523                      (setq al (cons key al))
 524                      ;; )
 525                      (setq value (cdr value)))
 526                    (insert-char-attributes ret
 527                                            readable
 528                                            (or al 'none) ; cal
 529                                            nil 'for-sub-node))
 530                (insert (prin1-to-string value)))
 531              (insert ")")
 532              (insert line-breaking))
 533             ((consp value)
 534              (insert (format "(%-18s " name))
 535              (setq lbs (concat "\n" (make-string (current-column) ?\ )))
 536              (while (consp value)
 537                (setq cell (car value))
 538                (if (and (consp cell)
 539                         (consp (car cell))
 540                         (setq ret (condition-case nil
 541                                       (find-char cell)
 542                                     (error nil)))
 543                         )
 544                    (progn
 545                      (setq rest cell
 546                            al nil
 547                            ;; cal nil
 548                            )
 549                      (while rest
 550                        (setq key (car (car rest)))
 551                        ;; (if (find-charset key)
 552                        ;;     (setq cal (cons key cal))
 553                        (setq al (cons key al))
 554                        ;; )
 555                        (setq rest (cdr rest)))
 556                      (if separator
 557                          (insert lbs))
 558                      (insert-char-attributes ret
 559                                              readable
 560                                              al ; cal
 561                                              nil 'for-sub-node)
 562                      (setq separator lbs))
 563                  (if separator
 564                      (insert separator))
 565                  (insert (prin1-to-string cell))
 566                  (setq separator " "))
 567                (setq value (cdr value)))
 568              (insert ")")
 569              (insert line-breaking))
 570             (t
 571              (insert (format "(%-18s . %S)%s"
 572                              name value
 573                              line-breaking))))
 574       (setq alist (cdr alist))))
 575   (insert ")"))
 576
 577 (defun char-db-insert-char-reference (plist &optional readable column)
 578   (unless column
 579     (setq column (current-column)))
 580   (let ((line-breaking
 581          (concat "\n" (make-string (1+ column) ?\ )))
 582         (separator "")
 583         name value)
 584     (insert "(")
 585     (while plist
 586       (setq name (pop plist))
 587       (setq value (pop plist))
 588       (cond ((eq name :char)
 589              (insert separator)
 590              (insert ":char\t")
 591              (cond ((numberp value)
 592                     (setq value (decode-char '=ucs value)))
 593                    ;; ((consp value)
 594                    ;;  (setq value (or (find-char value)
 595                    ;;                  value)))
 596                    )
 597              (char-db-insert-char-spec value readable)
 598              (insert line-breaking)
 599              (setq separator ""))
 600             ((eq name :radical)
 601              (insert (format "%s%s\t%d ; %c%s"
 602                              separator
 603                              name value
 604                              (ideographic-radical value)
 605                              line-breaking))
 606              (setq separator ""))
 607             (t
 608              (insert (format "%s%s\t%S" separator name value))
 609              (setq separator line-breaking)))
 610       ))
 611   (insert ")"))
 612
 613 (defun char-db-decode-isolated-char (ccs code-point)
 614   (let (ret)
 615     (setq ret
 616           (cond ((eq ccs 'arabic-iso8859-6)
 617                  (decode-char ccs code-point))
 618                 ;; ((eq ccs '=gt)
 619                 ;;  (decode-builtin-char '==gt code-point))
 620                 ((and (memq ccs '(=gt-pj-1
 621                                   =gt-pj-2
 622                                   =gt-pj-3
 623                                   =gt-pj-4
 624                                   =gt-pj-5
 625                                   =gt-pj-6
 626                                   =gt-pj-7
 627                                   =gt-pj-8
 628                                   =gt-pj-9
 629                                   =gt-pj-10
 630                                   =gt-pj-11))
 631                       (setq ret (decode-char ccs code-point))
 632                       (setq ret (encode-char ret '=gt 'defined-only)))
 633                  (decode-builtin-char '=gt ret))
 634                 (t
 635                  (decode-builtin-char ccs code-point))))
 636     (cond ((null ret)
 637            (or (decode-char ccs code-point)
 638                (define-char (list (cons ccs code-point)))))
 639           ((and (<= 0 (char-int ret))
 640                 (<= (char-int ret) #x1F))
 641            (decode-char '=ucs (+ #x2400 (char-int ret))))
 642           ((= (char-int ret) #x7F)
 643            ?\u2421)
 644           (t ret))))
 645
 646 (defvar char-db-convert-obsolete-format t)
 647
 648 (defun char-db-insert-ccs-feature (name value line-breaking)
 649   (cond
 650    ((integerp value)
 651     (insert
 652      (format
 653       (cond
 654        ((memq name '(=>iwds-1
 655                      ==shinjigen
 656                      =shinjigen
 657                      =shinjigen@1ed ==shinjigen@1ed
 658                      =shinjigen@rev ==shinjigen@rev
 659                      =shinjigen/+p@rev ==shinjigen/+p@rev
 660                      ===daikanwa/ho ==daikanwa/ho
 661                      =daikanwa/ho =>>daikanwa/ho =>daikanwa/ho))
 662         "(%-18s .  %04d)\t; %c")
 663        ((eq name '=shinjigen@1ed/24pr)
 664         "(%-18s . %04d)\t; %c")
 665        ((or
 666          (memq name
 667                '(===daikanwa
 668                  ==daikanwa =daikanwa =>>daikanwa =>daikanwa
 669                  =daikanwa@rev1 =daikanwa@rev2
 670                  =daikanwa/+p ==daikanwa/+p ===daikanwa/+p
 671                  =>>daikanwa/+p
 672                  =daikanwa/+2p =>>daikanwa/+2p
 673                  =gt ==gt ===gt
 674                  =>>gt =+>gt =>gt
 675                  =gt-k ==gt-k ===gt-k
 676                  =>>gt-k =>gt-k
 677                  =adobe-japan1-0 ==adobe-japan1-0 ===adobe-japan1-0
 678                  =adobe-japan1-1 ==adobe-japan1-1 ===adobe-japan1-1
 679                  =adobe-japan1-2 ==adobe-japan1-2 ===adobe-japan1-2
 680                  =adobe-japan1-3 ==adobe-japan1-3 ===adobe-japan1-3
 681                  =adobe-japan1-4 ==adobe-japan1-4 ===adobe-japan1-4
 682                  =adobe-japan1-5 ==adobe-japan1-5 ===adobe-japan1-5
 683                  =adobe-japan1-6 ==adobe-japan1-6 ===adobe-japan1-6
 684                  =>>adobe-japan1-0 =+>adobe-japan1-0
 685                  =>>adobe-japan1-1 =+>adobe-japan1-1
 686                  =>>adobe-japan1-2 =+>adobe-japan1-2
 687                  =>>adobe-japan1-3 =+>adobe-japan1-3
 688                  =>>adobe-japan1-4 =+>adobe-japan1-4
 689                  =>>adobe-japan1-5 =+>adobe-japan1-5
 690                  =>>adobe-japan1-6 =+>adobe-japan1-6
 691                  =>cbeta =cbeta =>>cbeta ==cbeta ===cbeta
 692                  =zinbun-oracle =>zinbun-oracle
 693                  ===hng-jou ===hng-keg ===hng-dng ===hng-mam
 694                  ===hng-drt ===hng-kgk ===hng-myz ===hng-kda
 695                  ===hng-khi ===hng-khm ===hng-hok ===hng-kyd ===hng-sok
 696                  ===hng-yhk ===hng-kak ===hng-kar ===hng-kae
 697                  ===hng-sys ===hng-tsu ===hng-tzj
 698                  ===hng-hos ===hng-nak ===hng-jhk
 699                  ===hng-hod ===hng-gok ===hng-ink ===hng-nto
 700                  ===hng-nkm ===hng-k24 ===hng-nkk
 701                  ===hng-kcc ===hng-kcj ===hng-kbk ===hng-sik
 702                  ===hng-skk ===hng-kyu ===hng-ksk ===hng-wan
 703                  ===hng-okd ===hng-wad ===hng-kmi
 704                  ===hng-zkd ===hng-doh ===hng-jyu
 705                  ===hng-tzs ===hng-kss ===hng-kyo
 706                  ===hng-smk))
 707          ;; (string-match "^=adobe-" (symbol-name name))
 708          )
 709         "(%-18s . %05d)\t; %c")
 710        ((memq name '(=hanyo-denshi/ks
 711                      ==hanyo-denshi/ks ===hanyo-denshi/ks
 712                      =>>hanyo-denshi/ks
 713                      =koseki ==koseki
 714                      =mj ==mj ===mj =>>mj
 715                      =zihai mojikyo))
 716         "(%-18s . %06d)\t; %c")
 717        ((>= (charset-dimension name) 2)
 718         "(%-18s . #x%04X)\t; %c")
 719        (t
 720         "(%-18s . #x%02X)\t; %c"))
 721       name
 722       (if (= (charset-iso-graphic-plane name) 1)
 723           (logior value
 724                   (cond ((= (charset-dimension name) 1)
 725                          #x80)
 726                         ((= (charset-dimension name) 2)
 727                          #x8080)
 728                         ((= (charset-dimension name) 3)
 729                          #x808080)
 730                         (t 0)))
 731         value)
 732       (char-db-decode-isolated-char name value)))
 733     (if (and (= (charset-chars name) 94)
 734              (= (charset-dimension name) 2))
 735         (insert (format " [%02d-%02d]"
 736                         (- (lsh value -8) 32)
 737                         (- (logand value 255) 32))))
 738     )
 739    (t
 740     (insert (format "(%-18s . %s)" name value))
 741     ))
 742   (insert line-breaking))
 743
 744 (defun char-db-insert-relation-feature (char name value line-breaking
 745                                              ccss readable)
 746   (insert (format "(%-18s%s " name line-breaking))
 747   (let ((lbs (concat "\n" (make-string (current-column) ?\ )))
 748         separator cell sources required-features
 749         ret)
 750     (while (consp value)
 751       (setq cell (car value))
 752       (if (integerp cell)
 753           (setq cell (decode-char '=ucs cell)))
 754       (cond
 755        ((eq name '->subsumptive)
 756         (when (or (not (some (lambda (atr)
 757                                (get-char-attribute cell atr))
 758                              char-db-ignored-attributes))
 759                   (some (lambda (ccs)
 760                           (encode-char cell ccs 'defined-only))
 761                         ccss))
 762           (if separator
 763               (insert lbs))
 764           (let ((char-db-ignored-attributes
 765                  (cons '<-subsumptive
 766                        char-db-ignored-attributes)))
 767             (insert-char-attributes cell readable nil nil 'for-sub-node))
 768           (setq separator lbs))
 769         )
 770        ((characterp cell)
 771         (setq sources
 772               (get-char-attribute
 773                char (intern (format "%s*sources" name))))
 774         (setq required-features nil)
 775         (dolist (source sources)
 776           (cond
 777            ((memq source '(JP
 778                            JP/Jouyou
 779                            shinjigen shinjigen@1ed shinjigen@rev))
 780             (setq required-features
 781                   (union required-features
 782                          '(=jis-x0208
 783                            =jis-x0208@1990
 784                            =jis-x0213-1@2000
 785                            =jis-x0213-1@2004
 786                            =jis-x0213-2
 787                            =jis-x0212
 788                            =jis-x0208@1983
 789                            =jis-x0208@1978
 790                            =shinjigen
 791                            =shinjigen@1ed
 792                            =shinjigen@rev
 793                            =shinjigen/+p@rev))))
 794            ((eq source 'CN)
 795             (setq required-features
 796                   (union required-features
 797                          '(=gb2312
 798                            =gb12345
 799                            =iso-ir165)))))
 800           (cond
 801            ((find-charset (setq ret (intern (format "=%s" source))))
 802             (setq required-features
 803                   (cons ret required-features)))
 804            (t (setq required-features
 805                     (cons source required-features)))))
 806         (cond ((string-match "@JP" (symbol-name name))
 807                (setq required-features
 808                      (union required-features
 809                             '(=jis-x0208
 810                               =jis-x0208@1990
 811                               =jis-x0213-1-2000
 812                               =jis-x0213-2-2000
 813                               =jis-x0212
 814                               =jis-x0208@1983
 815                               =jis-x0208@1978))))
 816               ((string-match "@CN" (symbol-name name))
 817                (setq required-features
 818                      (union required-features
 819                             '(=gb2312
 820                               =gb12345
 821                               =iso-ir165)))))
 822         (if separator
 823             (insert lbs))
 824         (if readable
 825             (insert (format "%S" cell))
 826           (char-db-insert-char-spec cell readable
 827                                     nil
 828                                     required-features))
 829         (setq separator lbs))
 830        ((consp cell)
 831         (if separator
 832             (insert lbs))
 833         (if (consp (car cell))
 834             (char-db-insert-char-spec cell readable)
 835           (char-db-insert-char-reference cell readable))
 836         (setq separator lbs))
 837        (t
 838         (if separator
 839             (insert separator))
 840         (insert (prin1-to-string cell))
 841         (setq separator " ")))
 842       (setq value (cdr value)))
 843     (insert ")")
 844     (insert line-breaking)))
 845
 846 (defun insert-char-attributes (char &optional readable attributes column
 847                                     for-sub-node)
 848   (unless column
 849     (setq column (current-column)))
 850   (let (name value ; has-long-ccs-name
 851         rest
 852         radical strokes
 853         (line-breaking
 854          (concat "\n" (make-string (1+ column) ?\ )))
 855         lbs cell separator ret
 856         key al cal
 857         dest-ccss ; sources required-features
 858         ccss)
 859     (let (atr-d)
 860       (setq attributes
 861             (sort (if attributes
 862                       (if (consp attributes)
 863                           (progn
 864                             (dolist (name attributes)
 865                               (unless (memq name char-db-ignored-attributes)
 866                                 (if (find-charset name)
 867                                     (push name ccss))
 868                                 (push name atr-d)))
 869                             atr-d))
 870                     (dolist (name (char-attribute-list))
 871                       (unless (memq name char-db-ignored-attributes)
 872                         (if (find-charset name)
 873                             (push name ccss))
 874                         (push name atr-d)))
 875                     atr-d)
 876                   #'char-attribute-name<)))
 877     (insert "(")
 878     (when (memq '<-subsumptive attributes)
 879       (when (or readable (not for-sub-node))
 880         (when (setq value (get-char-attribute char '<-subsumptive))
 881           (char-db-insert-relation-feature char '<-subsumptive value
 882                                            line-breaking
 883                                            ccss readable)))
 884       (setq attributes (delq '<-subsumptive attributes)))
 885     (when (and (memq '<-denotational attributes)
 886                (setq value (get-char-attribute char '<-denotational)))
 887       (char-db-insert-relation-feature char '<-denotational value
 888                                        line-breaking
 889                                        ccss readable)
 890       (setq attributes (delq '<-denotational attributes)))
 891     (when (and (memq '<-denotational@component attributes)
 892                (setq value (get-char-attribute char '<-denotational@component)))
 893       (char-db-insert-relation-feature char '<-denotational@component value
 894                                        line-breaking
 895                                        ccss readable)
 896       (setq attributes (delq '<-denotational@component attributes)))
 897     (when (and (memq 'name attributes)
 898                (setq value (get-char-attribute char 'name)))
 899       (insert (format
 900                (if (> (+ (current-column) (length value)) 48)
 901                    "(name . %S)%s"
 902                  "(name               . %S)%s")
 903                value line-breaking))
 904       (setq attributes (delq 'name attributes))
 905       )
 906     (when (and (memq 'name* attributes)
 907                (setq value (get-char-attribute char 'name*)))
 908       (insert (format
 909                (if (> (+ (current-column) (length value)) 48)
 910                    "(name* . %S)%s"
 911                  "(name*              . %S)%s")
 912                value line-breaking))
 913       (setq attributes (delq 'name* attributes))
 914       )
 915     (when (and (memq 'script attributes)
 916                (setq value (get-char-attribute char 'script)))
 917       (insert (format "(script\t\t%s)%s"
 918                       (mapconcat (function prin1-to-string)
 919                                  value " ")
 920                       line-breaking))
 921       (setq attributes (delq 'script attributes))
 922       )
 923     (dolist (name '(=>ucs =>ucs*))
 924       (when (and (memq name attributes)
 925                  (setq value (get-char-attribute char name)))
 926         (insert (format "(%-18s . #x%04X)\t; %c%s"
 927                         name value (decode-char '=ucs value)
 928                         line-breaking))
 929         (setq attributes (delq name attributes))))
 930     (dolist (name '(=>ucs@gb =>ucs@big5))
 931       (when (and (memq name attributes)
 932                  (setq value (get-char-attribute char name)))
 933         (insert (format "(%-18s . #x%04X)\t; %c%s"
 934                         name value
 935                         (decode-char (intern
 936                                       (concat "="
 937                                               (substring
 938                                                (symbol-name name) 2)))
 939                                      value)
 940                         line-breaking))
 941         (setq attributes (delq name attributes))
 942         ))
 943     ;; (dolist (name '(=>daikanwa))
 944     ;;   (when (and (memq name attributes)
 945     ;;              (setq value (get-char-attribute char name)))
 946     ;;     (insert
 947     ;;      (if (integerp value)
 948     ;;          (format "(%-18s . %05d)\t; %c%s"
 949     ;;                  name value (decode-char '=daikanwa value)
 950     ;;                  line-breaking)
 951     ;;        (format "(%-18s %s)\t; %c%s"
 952     ;;                name
 953     ;;                (mapconcat (function prin1-to-string)
 954     ;;                           value " ")
 955     ;;                (char-representative-of-daikanwa char)
 956     ;;                line-breaking)))
 957     ;;     (setq attributes (delq name attributes))))
 958     (when (and (memq 'general-category attributes)
 959                (setq value (get-char-attribute char 'general-category)))
 960       (insert (format
 961                "(general-category\t%s) ; %s%s"
 962                (mapconcat (lambda (cell)
 963                             (format "%S" cell))
 964                           value " ")
 965                (cond ((rassoc value unidata-normative-category-alist)
 966                       "Normative Category")
 967                      ((rassoc value unidata-informative-category-alist)
 968                       "Informative Category")
 969                      (t
 970                       "Unknown Category"))
 971                line-breaking))
 972       (setq attributes (delq 'general-category attributes))
 973       )
 974     (when (and (memq 'bidi-category attributes)
 975                (setq value (get-char-attribute char 'bidi-category)))
 976       (insert (format "(bidi-category\t. %S)%s"
 977                       value
 978                       line-breaking))
 979       (setq attributes (delq 'bidi-category attributes))
 980       )
 981     (unless (or (not (memq 'mirrored attributes))
 982                 (eq (setq value (get-char-attribute char 'mirrored 'empty))
 983                     'empty))
 984       (insert (format "(mirrored\t\t. %S)%s"
 985                       value
 986                       line-breaking))
 987       (setq attributes (delq 'mirrored attributes))
 988       )
 989     (cond
 990      ((and (memq 'decimal-digit-value attributes)
 991            (setq value (get-char-attribute char 'decimal-digit-value)))
 992       (insert (format "(decimal-digit-value . %S)%s"
 993                       value
 994                       line-breaking))
 995       (setq attributes (delq 'decimal-digit-value attributes))
 996       (when (and (memq 'digit-value attributes)
 997                  (setq value (get-char-attribute char 'digit-value)))
 998         (insert (format "(digit-value\t . %S)%s"
 999                         value
1000                         line-breaking))
1001         (setq attributes (delq 'digit-value attributes))
1002         )
1003       (when (and (memq 'numeric-value attributes)
1004                  (setq value (get-char-attribute char 'numeric-value)))
1005         (insert (format "(numeric-value\t . %S)%s"
1006                         value
1007                         line-breaking))
1008         (setq attributes (delq 'numeric-value attributes))
1009         )
1010       )
1011      (t
1012       (when (and (memq 'digit-value attributes)
1013                  (setq value (get-char-attribute char 'digit-value)))
1014         (insert (format "(digit-value\t. %S)%s"
1015                         value
1016                         line-breaking))
1017         (setq attributes (delq 'digit-value attributes))
1018         )
1019       (when (and (memq 'numeric-value attributes)
1020                  (setq value (get-char-attribute char 'numeric-value)))
1021         (insert (format "(numeric-value\t. %S)%s"
1022                         value
1023                         line-breaking))
1024         (setq attributes (delq 'numeric-value attributes))
1025         )))
1026     (when (and (memq 'iso-10646-comment attributes)
1027                (setq value (get-char-attribute char 'iso-10646-comment)))
1028       (insert (format "(iso-10646-comment\t. %S)%s"
1029                       value
1030                       line-breaking))
1031       (setq attributes (delq 'iso-10646-comment attributes))
1032       )
1033     (when (and (memq 'morohashi-daikanwa attributes)
1034                (setq value (get-char-attribute char 'morohashi-daikanwa)))
1035       (insert (format "(morohashi-daikanwa\t%s)%s"
1036                       (mapconcat (function prin1-to-string) value " ")
1037                       line-breaking))
1038       (setq attributes (delq 'morohashi-daikanwa attributes))
1039       )
1040     (setq radical nil
1041           strokes nil)
1042     (when (and (memq 'ideographic-radical attributes)
1043                (setq value (get-char-attribute char 'ideographic-radical)))
1044       (setq radical value)
1045       (insert (format "(ideographic-radical . %S)\t; %c%s"
1046                       radical
1047                       (ideographic-radical radical)
1048                       line-breaking))
1049       (setq attributes (delq 'ideographic-radical attributes))
1050       )
1051     (when (and (memq 'shuowen-radical attributes)
1052                (setq value (get-char-attribute char 'shuowen-radical)))
1053       (insert (format "(shuowen-radical\t. %S)\t; %c%s"
1054                       value
1055                       (shuowen-radical value)
1056                       line-breaking))
1057       (setq attributes (delq 'shuowen-radical attributes))
1058       )
1059     (let (key)
1060       (dolist (domain
1061                (append
1062                 char-db-feature-domains
1063                 (let (dest domain)
1064                   (dolist (feature (char-attribute-list))
1065                     (setq feature (symbol-name feature))
1066                     (when (string-match
1067                            "\\(radical\\|strokes\\)@\\([^@*]+\\)\\(\\*\\|$\\)"
1068                            feature)
1069                       (setq domain (intern (match-string 2 feature)))
1070                      (unless (memq domain dest)
1071                        (setq dest (cons domain dest)))))
1072                   (sort dest #'string<))))
1073         (setq key (intern (format "%s@%s" 'ideographic-radical domain)))
1074         (when (and (memq key attributes)
1075                    (setq value (get-char-attribute char key)))
1076           (setq radical value)
1077           (insert (format "(%s . %S)\t; %c%s"
1078                           key
1079                           radical
1080                           (ideographic-radical radical)
1081                           line-breaking))
1082           (setq attributes (delq key attributes))
1083           )
1084         (setq key (intern (format "%s@%s" 'ideographic-strokes domain)))
1085         (when (and (memq key attributes)
1086                    (setq value (get-char-attribute char key)))
1087           (setq strokes value)
1088           (insert (format "(%s . %S)%s"
1089                           key
1090                           strokes
1091                           line-breaking))
1092           (setq attributes (delq key attributes))
1093           )
1094         (setq key (intern (format "%s@%s" 'total-strokes domain)))
1095         (when (and (memq key attributes)
1096                    (setq value (get-char-attribute char key)))
1097           (insert (format "(%s       . %S)%s"
1098                           key
1099                           value
1100                           line-breaking))
1101           (setq attributes (delq key attributes))
1102           )
1103         (dolist (feature '(ideographic-radical
1104                            ideographic-strokes
1105                            total-strokes))
1106           (setq key (intern (format "%s@%s*sources" feature domain)))
1107           (when (and (memq key attributes)
1108                      (setq value (get-char-attribute char key)))
1109             (insert (format "(%s%s" key line-breaking))
1110             (dolist (cell value)
1111               (insert (format " %s" cell)))
1112             (insert ")")
1113             (insert line-breaking)
1114             (setq attributes (delq key attributes))
1115             ))
1116         ))
1117     (when (and (memq 'ideographic-strokes attributes)
1118                (setq value (get-char-attribute char 'ideographic-strokes)))
1119       (setq strokes value)
1120       (insert (format "(ideographic-strokes . %S)%s"
1121                       strokes
1122                       line-breaking))
1123       (setq attributes (delq 'ideographic-strokes attributes))
1124       )
1125     (when (and (memq 'kangxi-radical attributes)
1126                (setq value (get-char-attribute char 'kangxi-radical)))
1127       (unless (eq value radical)
1128         (insert (format "(kangxi-radical\t . %S)\t; %c%s"
1129                         value
1130                         (ideographic-radical value)
1131                         line-breaking))
1132         (or radical
1133             (setq radical value)))
1134       (setq attributes (delq 'kangxi-radical attributes))
1135       )
1136     (when (and (memq 'kangxi-strokes attributes)
1137                (setq value (get-char-attribute char 'kangxi-strokes)))
1138       (unless (eq value strokes)
1139         (insert (format "(kangxi-strokes\t . %S)%s"
1140                         value
1141                         line-breaking))
1142         (or strokes
1143             (setq strokes value)))
1144       (setq attributes (delq 'kangxi-strokes attributes))
1145       )
1146     (when (and (memq 'japanese-radical attributes)
1147                (setq value (get-char-attribute char 'japanese-radical)))
1148       (unless (eq value radical)
1149         (insert (format "(japanese-radical\t . %S)\t; %c%s"
1150                         value
1151                         (ideographic-radical value)
1152                         line-breaking))
1153         (or radical
1154             (setq radical value)))
1155       (setq attributes (delq 'japanese-radical attributes))
1156       )
1157     (when (and (memq 'japanese-strokes attributes)
1158                (setq value (get-char-attribute char 'japanese-strokes)))
1159       (unless (eq value strokes)
1160         (insert (format "(japanese-strokes\t . %S)%s"
1161                         value
1162                         line-breaking))
1163         (or strokes
1164             (setq strokes value)))
1165       (setq attributes (delq 'japanese-strokes attributes))
1166       )
1167     (when (and (memq 'cns-radical attributes)
1168                (setq value (get-char-attribute char 'cns-radical)))
1169       (insert (format "(cns-radical\t . %S)\t; %c%s"
1170                       value
1171                       (ideographic-radical value)
1172                       line-breaking))
1173       (setq attributes (delq 'cns-radical attributes))
1174       )
1175     (when (and (memq 'cns-strokes attributes)
1176                (setq value (get-char-attribute char 'cns-strokes)))
1177       (unless (eq value strokes)
1178         (insert (format "(cns-strokes\t . %S)%s"
1179                         value
1180                         line-breaking))
1181         (or strokes
1182             (setq strokes value)))
1183       (setq attributes (delq 'cns-strokes attributes))
1184       )
1185     ;; (when (and (memq 'shinjigen-1-radical attributes)
1186     ;;            (setq value (get-char-attribute char 'shinjigen-1-radical)))
1187     ;;   (unless (eq value radical)
1188     ;;     (insert (format "(shinjigen-1-radical . %S)\t; %c%s"
1189     ;;                     value
1190     ;;                     (ideographic-radical value)
1191     ;;                     line-breaking))
1192     ;;     (or radical
1193     ;;         (setq radical value)))
1194     ;;   (setq attributes (delq 'shinjigen-1-radical attributes))
1195     ;;   )
1196     (when (and (memq 'ideographic- attributes)
1197                (setq value (get-char-attribute char 'ideographic-)))
1198       (insert "(ideographic-       ")
1199       (setq lbs (concat "\n" (make-string (current-column) ?\ ))
1200             separator nil)
1201       (while (consp value)
1202         (setq cell (car value))
1203         (if (integerp cell)
1204             (setq cell (decode-char '=ucs cell)))
1205         (cond ((characterp cell)
1206                (if separator
1207                    (insert lbs))
1208                (if readable
1209                    (insert (format "%S" cell))
1210                  (char-db-insert-char-spec cell readable))
1211                (setq separator lbs))
1212               ((consp cell)
1213                (if separator
1214                    (insert lbs))
1215                (if (consp (car cell))
1216                    (char-db-insert-char-spec cell readable)
1217                  (char-db-insert-char-reference cell readable))
1218                (setq separator lbs))
1219               (t
1220                (if separator
1221                    (insert separator))
1222                (insert (prin1-to-string cell))
1223                (setq separator " ")))
1224         (setq value (cdr value)))
1225       (insert ")")
1226       (insert line-breaking)
1227       (setq attributes (delq 'ideographic- attributes)))
1228     (when (and (memq 'total-strokes attributes)
1229                (setq value (get-char-attribute char 'total-strokes)))
1230       (insert (format "(total-strokes       . %S)%s"
1231                       value
1232                       line-breaking))
1233       (setq attributes (delq 'total-strokes attributes))
1234       )
1235     (when (and (memq '->ideograph attributes)
1236                (setq value (get-char-attribute char '->ideograph)))
1237       (insert (format "(->ideograph\t%s)%s"
1238                       (mapconcat (lambda (code)
1239                                    (cond ((symbolp code)
1240                                           (symbol-name code))
1241                                          ((integerp code)
1242                                           (format "#x%04X" code))
1243                                          (t
1244                                           (format "%s %S"
1245                                                   line-breaking code))))
1246                                  value " ")
1247                       line-breaking))
1248       (setq attributes (delq '->ideograph attributes))
1249       )
1250     ;; (when (and (memq '->decomposition attributes)
1251     ;;            (setq value (get-char-attribute char '->decomposition)))
1252     ;;   (insert (format "(->decomposition\t%s)%s"
1253     ;;                   (mapconcat (lambda (code)
1254     ;;                                (cond ((symbolp code)
1255     ;;                                       (symbol-name code))
1256     ;;                                      ((characterp code)
1257     ;;                                       (if readable
1258     ;;                                           (format "%S" code)
1259     ;;                                         (format "#x%04X"
1260     ;;                                                 (char-int code))
1261     ;;                                         ))
1262     ;;                                      ((integerp code)
1263     ;;                                       (format "#x%04X" code))
1264     ;;                                      (t
1265     ;;                                       (format "%s%S" line-breaking code))))
1266     ;;                              value " ")
1267     ;;                   line-breaking))
1268     ;;   (setq attributes (delq '->decomposition attributes))
1269     ;;   )
1270     (if (equal (get-char-attribute char '->titlecase)
1271                (get-char-attribute char '->uppercase))
1272         (setq attributes (delq '->titlecase attributes)))
1273     (when (and (memq '->mojikyo attributes)
1274                (setq value (get-char-attribute char '->mojikyo)))
1275       (insert (format "(->mojikyo\t\t. %06d)\t; %c%s"
1276                       value (decode-char 'mojikyo value)
1277                       line-breaking))
1278       (setq attributes (delq '->mojikyo attributes))
1279       )
1280     (when (and (memq 'hanyu-dazidian-vol attributes)
1281                (setq value (get-char-attribute char 'hanyu-dazidian-vol)))
1282       (insert (format "(hanyu-dazidian-vol  . %d)%s"
1283                       value line-breaking))
1284       (setq attributes (delq 'hanyu-dazidian-vol attributes))
1285       )
1286     (when (and (memq 'hanyu-dazidian-page attributes)
1287                (setq value (get-char-attribute char 'hanyu-dazidian-page)))
1288       (insert (format "(hanyu-dazidian-page . %d)%s"
1289                       value line-breaking))
1290       (setq attributes (delq 'hanyu-dazidian-page attributes))
1291       )
1292     (when (and (memq 'hanyu-dazidian-char attributes)
1293                (setq value (get-char-attribute char 'hanyu-dazidian-char)))
1294       (insert (format "(hanyu-dazidian-char . %d)%s"
1295                       value line-breaking))
1296       (setq attributes (delq 'hanyu-dazidian-char attributes))
1297       )
1298     (unless readable
1299       (dolist (ignored '(composition
1300                          ->denotational <-subsumptive ->ucs-unified
1301                          ->ideographic-component-forms))
1302         (setq attributes (delq ignored attributes))))
1303     (while attributes
1304       (setq name (car attributes))
1305       (unless (eq (setq value (get-char-attribute char name 'value-is-empty))
1306                   'value-is-empty)
1307         (cond ((setq ret (find-charset name))
1308                (setq name (charset-name ret))
1309                (when (not (memq name dest-ccss))
1310                  (setq dest-ccss (cons name dest-ccss))
1311                  (char-db-insert-ccs-feature name value line-breaking))
1312                )
1313               ((string-match "^=>ucs@" (symbol-name name))
1314                (insert (format "(%-18s . #x%04X)\t; %c%s"
1315                                name value (decode-char '=ucs value)
1316                                line-breaking))
1317                )
1318               ((eq name 'jisx0208-1978/4X)
1319                (insert (format "(%-18s . #x%04X)%s"
1320                                name value
1321                                line-breaking))
1322                )
1323               ((and
1324                 (not readable)
1325                 (not (eq name '->subsumptive))
1326                 (not (eq name '->uppercase))
1327                 (not (eq name '->lowercase))
1328                 (not (eq name '->titlecase))
1329                 (not (eq name '->canonical))
1330                 (not (eq name '->Bopomofo))
1331                 (not (eq name '->mistakable))
1332                 (not (eq name '->ideographic-variants))
1333                 (null (get-char-attribute
1334                        char (intern (format "%s*sources" name))))
1335                 (not (string-match "\\*sources$" (symbol-name name)))
1336                 (null (get-char-attribute
1337                        char (intern (format "%s*note" name))))
1338                 (not (string-match "\\*note$" (symbol-name name)))
1339                 (or (eq name '<-identical)
1340                     (eq name '<-uppercase)
1341                     (eq name '<-lowercase)
1342                     (eq name '<-titlecase)
1343                     (eq name '<-canonical)
1344                     (eq name '<-ideographic-variants)
1345                     ;; (eq name '<-synonyms)
1346                     (string-match "^<-synonyms" (symbol-name name))
1347                     (eq name '<-mistakable)
1348                     (when (string-match "^->" (symbol-name name))
1349                       (cond
1350                        ((string-match "^->fullwidth" (symbol-name name))
1351                         (not (and (consp value)
1352                                   (characterp (car value))
1353                                   (encode-char
1354                                    (car value) '=ucs 'defined-only)))
1355                         )
1356                        (t)))
1357                     ))
1358                )
1359               ((or (eq name 'ideographic-structure)
1360                    (eq name 'ideographic-combination)
1361                    (eq name 'ideographic-)
1362                    (eq name '=decomposition)
1363                    (char-feature-base-name= '=decomposition name)
1364                    (char-feature-base-name= '=>decomposition name)
1365                    ;; (string-match "^=>*decomposition\\(@[^*]+\\)?$"
1366                    ;;               (symbol-name name))
1367                    (string-match "^\\(->\\|<-\\)[^*]*$" (symbol-name name))
1368                    (string-match "^\\(->\\|<-\\)[^*]*\\*sources$"
1369                                  (symbol-name name))
1370                    )
1371                (char-db-insert-relation-feature char name value
1372                                                 line-breaking
1373                                                 ccss readable))
1374               ((memq name '(ideograph=
1375                             original-ideograph-of
1376                             ancient-ideograph-of
1377                             vulgar-ideograph-of
1378                             wrong-ideograph-of
1379                             ;; simplified-ideograph-of
1380                             ideographic-variants
1381                             ;; ideographic-different-form-of
1382                             ))
1383                (insert (format "(%-18s%s " name line-breaking))
1384                (setq lbs (concat "\n" (make-string (current-column) ?\ ))
1385                      separator nil)
1386                (while (consp value)
1387                  (setq cell (car value))
1388                  (if (and (consp cell)
1389                           (consp (car cell)))
1390                      (progn
1391                        (if separator
1392                            (insert lbs))
1393                        (char-db-insert-alist cell readable)
1394                        (setq separator lbs))
1395                    (if separator
1396                        (insert separator))
1397                    (insert (prin1-to-string cell))
1398                    (setq separator " "))
1399                  (setq value (cdr value)))
1400                (insert ")")
1401                (insert line-breaking))
1402               ((consp value)
1403                (insert (format "(%-18s " name))
1404                (setq lbs (concat "\n" (make-string (current-column) ?\ ))
1405                      separator nil)
1406                (while (consp value)
1407                  (setq cell (car value))
1408                  (if (and (consp cell)
1409                           (consp (car cell))
1410                           (setq ret (condition-case nil
1411                                         (find-char cell)
1412                                       (error nil))))
1413                      (progn
1414                        (setq rest cell
1415                              al nil
1416                              cal nil)
1417                        (while rest
1418                          (setq key (car (car rest)))
1419                          (if (find-charset key)
1420                              (setq cal (cons key cal))
1421                            (setq al (cons key al)))
1422                          (setq rest (cdr rest)))
1423                        (if separator
1424                            (insert lbs))
1425                        (insert-char-attributes ret
1426                                                readable
1427                                                al ; cal
1428                                                nil 'for-sub-node)
1429                        (setq separator lbs))
1430                    (setq ret (prin1-to-string cell))
1431                    (if separator
1432                        (if (< (+ (current-column)
1433                                  (length ret)
1434                                  (length separator))
1435                               76)
1436                            (insert separator)
1437                          (insert lbs)))
1438                    (insert ret)
1439                    (setq separator " "))
1440                  (setq value (cdr value)))
1441                (insert ")")
1442                (insert line-breaking))
1443               (t
1444                (insert (format "(%-18s" name))
1445                (setq ret (prin1-to-string value))
1446                (unless (< (+ (current-column)
1447                              (length ret)
1448                              3)
1449                           76)
1450                  (insert line-breaking))
1451                (insert " . " ret ")" line-breaking)
1452                ;; (insert (format "(%-18s . %S)%s"
1453                ;;                 name value
1454                ;;                 line-breaking))
1455                )
1456               ))
1457       (setq attributes (cdr attributes)))
1458     (insert ")")))
1459
1460 (defun insert-char-data (char &optional readable
1461                               attributes)
1462   (save-restriction
1463     (narrow-to-region (point)(point))
1464     (insert "(define-char
1465   '")
1466     (insert-char-attributes char readable attributes)
1467     (insert ")\n")
1468     (goto-char (point-min))
1469     (while (re-search-forward "[ \t]+$" nil t)
1470       (replace-match ""))
1471     ;; from tabify.
1472     (goto-char (point-min))
1473     (while (re-search-forward "[ \t][ \t][ \t]*" nil t)
1474       (let ((column (current-column))
1475             (indent-tabs-mode t))
1476         (delete-region (match-beginning 0) (point))
1477         (indent-to column)))
1478     (goto-char (point-max))
1479     ;; (tabify (point-min)(point-max))
1480     ))
1481
1482 (defun insert-char-data-with-variant (char &optional printable
1483                                            no-ucs-unified
1484                                            script excluded-script)
1485   (insert-char-data char printable)
1486   (let ((variants (char-variants char))
1487         rest
1488         variant vs ret)
1489     (setq variants (sort variants #'<))
1490     (setq rest variants)
1491     (setq variants (cons char variants))
1492     (while rest
1493       (setq variant (car rest))
1494       (unless (get-char-attribute variant '<-subsumptive)
1495         (if (and (or (null script)
1496                      (null (setq vs (get-char-attribute variant 'script)))
1497                      (memq script vs))
1498                  (or (null excluded-script)
1499                      (null (setq vs (get-char-attribute variant 'script)))
1500                      (not (memq excluded-script vs))))
1501             (unless (and no-ucs-unified (get-char-attribute variant '=ucs))
1502               (insert-char-data variant printable)
1503               (if (setq ret (char-variants variant))
1504                   (while ret
1505                     (or (memq (car ret) variants)
1506                         ;; (get-char-attribute (car ret) '<-subsumptive)
1507                         (setq rest (nconc rest (list (car ret)))))
1508                     (setq ret (cdr ret)))))))
1509       (setq rest (cdr rest)))))
1510
1511 (defun insert-char-range-data (min max &optional script excluded-script)
1512   (let ((code min)
1513         char)
1514     (while (<= code max)
1515       (setq char (decode-char '=ucs code))
1516       (if (encode-char char '=ucs 'defined-only)
1517           (insert-char-data-with-variant char nil 'no-ucs-unified
1518                                          script excluded-script))
1519       (setq code (1+ code)))))
1520
1521 (defun write-char-range-data-to-file (min max file
1522                                           &optional script excluded-script)
1523   (let ((coding-system-for-write char-db-file-coding-system))
1524     (with-temp-buffer
1525       (insert (format ";; -*- coding: %s -*-\n"
1526                       char-db-file-coding-system))
1527       (insert-char-range-data min max script excluded-script)
1528       (write-region (point-min)(point-max) file))))
1529
1530 (defvar what-character-original-window-configuration)
1531
1532 ;;;###autoload
1533 (defun what-char-definition (char)
1534   (interactive (list (char-after)))
1535   (let ((buf (get-buffer-create "*Character Description*"))
1536         (the-buf (current-buffer))
1537         (win-conf (current-window-configuration)))
1538     (pop-to-buffer buf)
1539     (make-local-variable 'what-character-original-window-configuration)
1540     (setq what-character-original-window-configuration win-conf)
1541     (setq buffer-read-only nil)
1542     (erase-buffer)
1543     (condition-case err
1544         (progn
1545           (insert-char-data-with-variant char 'printable)
1546           (unless (char-attribute-alist char)
1547             (insert (format ";; = %c\n"
1548                             (let* ((rest (split-char char))
1549                                    (ccs (pop rest))
1550                                    (code (pop rest)))
1551                               (while rest
1552                                 (setq code (logior (lsh code 8)
1553                                                    (pop rest))))
1554                               (decode-char ccs code)))))
1555           ;; (char-db-update-comment)
1556           (set-buffer-modified-p nil)
1557           (view-mode the-buf (lambda (buf)
1558                                (set-window-configuration
1559                                 what-character-original-window-configuration)
1560                                ))
1561           (goto-char (point-min)))
1562       (error (progn
1563                (set-window-configuration
1564                 what-character-original-window-configuration)
1565                (signal (car err) (cdr err)))))))
1566
1567
1568 ;;; @ end
1569 ;;;
1570
1571 (provide 'char-db-util)
1572
1573 ;;; char-db-util.el ends here