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