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