git.chise.org Git - chise/xemacs-chise.git.1/blob - lisp/utf-2000/char-db-util.el

   1 ;;; char-db-util.el --- Character Database utility -*- coding: utf-8-er; -*-
   2
   3 ;; Copyright (C) 1998,1999,2000,2001,2002,2003,2004,2005,2006,2007,
   4 ;;   2008,2009,2010,2011,2012,2013,2014,2015,2016 MORIOKA Tomohiko.
   5
   6 ;; Author: MORIOKA Tomohiko <tomo@kanji.zinbun.kyoto-u.ac.jp>
   7 ;; Keywords: CHISE, Character Database, ISO/IEC 10646, UCS, Unicode, MULE.
   8
   9 ;; This file is part of XEmacs CHISE.
  10
  11 ;; XEmacs CHISE is free software; you can redistribute it and/or
  12 ;; modify it under the terms of the GNU General Public License as
  13 ;; published by the Free Software Foundation; either version 2, or (at
  14 ;; your option) any later version.
  15
  16 ;; XEmacs CHISE is distributed in the hope that it will be useful,
  17 ;; but WITHOUT ANY WARRANTY; without even the implied warranty of
  18 ;; MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the GNU
  19 ;; General Public License for more details.
  20
  21 ;; You should have received a copy of the GNU General Public License
  22 ;; along with XEmacs CHISE; see the file COPYING.  If not, write to
  23 ;; the Free Software Foundation, Inc., 59 Temple Place - Suite 330,
  24 ;; Boston, MA 02111-1307, USA.
  25
  26 ;;; Code:
  27
  28 (require 'chise-subr)
  29 (require 'ideograph-subr)
  30
  31 (defconst unidata-normative-category-alist
  32   '(("Lu" letter        uppercase)
  33     ("Ll" letter        lowercase)
  34     ("Lt" letter        titlecase)
  35     ("Mn" mark          non-spacing)
  36     ("Mc" mark          spacing-combining)
  37     ("Me" mark          enclosing)
  38     ("Nd" number        decimal-digit)
  39     ("Nl" number        letter)
  40     ("No" number        other)
  41     ("Zs" separator     space)
  42     ("Zl" separator     line)
  43     ("Zp" separator     paragraph)
  44     ("Cc" other         control)
  45     ("Cf" other         format)
  46     ("Cs" other         surrogate)
  47     ("Co" other         private-use)
  48     ("Cn" other         not-assigned)))
  49
  50 (defconst unidata-informative-category-alist
  51   '(("Lm" letter        modifier)
  52     ("Lo" letter        other)
  53     ("Pc" punctuation   connector)
  54     ("Pd" punctuation   dash)
  55     ("Ps" punctuation   open)
  56     ("Pe" punctuation   close)
  57     ("Pi" punctuation   initial-quote)
  58     ("Pf" punctuation   final-quote)
  59     ("Po" punctuation   other)
  60     ("Sm" symbol        math)
  61     ("Sc" symbol        currency)
  62     ("Sk" symbol        modifier)
  63     ("So" symbol        other)
  64     ))
  65
  66 (defconst shuowen-radicals
  67   [?一 ?上 ?示 ?三 ?王 ?玉 ?玨 ?气 ?士 ?丨 ; 010
  68    ?屮 ?艸 ?蓐 ?茻 ?小 ?八 ?釆 ?半 ?牛 ?犛 ; 020
  69    ?告 ?口 ?凵 ?吅 ?哭 ?走 ?止 ?癶 ?步 ?此 ; 030
  70    ?正 ?是 ?辵 ?彳 ?廴 ?㢟 ?行 ?齒 ?牙 ?足 ; 040
  71    ?疋 ?品 ?龠 ?冊 ?㗊 ?舌 ?干 ?谷 ?只 ?㕯 ; 050
  72    ?句 ?丩 ?古 ?十 ?卅 ?言 ?誩 ?音 ?䇂 ?丵 ; 060
  73    ?菐 ?𠬞 ?𠬜 ?共 ?異 ?舁 ?𦥑 ?䢅 ?爨 ?革 ; 070
  74    ?鬲 ?䰜 ?爪 ?𠃨 ?鬥 ?又 ?𠂇 ?㕜 ?支 ?𦘒 ; 080
  75    ?聿 ?畫 ?隶 ?臤 ?臣 ?殳 ?殺 ?𠘧 ?寸 ?皮 ; 090
  76    ?㼱 ?攴 ?敎 ?卜 ?用 ?爻 ?㸚 ?𥄎 ?目 ?䀠 ; 100
  77    ?眉 ?盾 ?自 ?白 ?鼻 ?皕 ?習 ?羽 ?隹 ?奞 ; 110
  78    ?萑 ?𦫳 ?苜 ?羊 ?羴 ?瞿 ?雔 ?雥 ?鳥 ?烏 ; 120
  79    ?𠦒 ?冓 ?幺 ?𢆶 ?叀 ?玄 ?予 ?放 ?𠬪 ?𣦼 ; 130
  80    ?歺 ?死 ?冎 ?骨 ?肉 ?筋 ?刀 ?刃 ?㓞 ?丰 ; 140
  81    ?耒 ?𧢲 ?竹 ?箕 ?丌 ?左 ?工 ?㠭 ?巫 ?甘 ; 150
  82    ?曰 ?乃 ?丂 ?可 ?兮 ?号 ?亏 ?旨 ?喜 ?壴 ; 160
  83 ;  ?旨 ?曰 ?乃 ?丂 ?可 ?兮 ?号 ?亏 ?喜 ?壴 ; 160
  84    ?鼓 ?豈 ?豆 ?豊 ?豐 ?䖒 ?虍 ?虎 ?虤 ?皿 ; 170
  85    ?𠙴 ?去 ?血 ?丶 ?丹 ?青 ?井 ?皀 ?鬯 ?食 ; 180
  86    ?亼 ?會 ?倉 ?入 ?缶 ?矢 ?高 ?冂 ?𩫏 ?京 ; 190
  87    ?亯 ?𣆪 ?畗 ?㐭 ?嗇 ?來 ?麥 ?夊 ?舛 ?䑞 ; 200
  88    ?韋 ?弟 ?夂 ?久 ?桀 ?木 ?東 ?林 ?才 ?叒 ; 210
  89    ?之 ?帀 ?出 ?𣎵 ?生 ?乇 ?𠂹 ?𠌶 ?華 ?𥝌 ; 220
  90    ?稽 ?巢 ?桼 ?束 ?㯻 ?囗 ?員 ?貝 ?邑 ?𨛜 ; 230
  91    ?日 ?旦 ?倝 ?㫃 ?冥 ?晶 ?月 ?有 ?明 ?囧 ; 240
  92    ?夕 ?多 ?毌 ?𢎘 ?𣐺 ?卣 ?齊 ?朿 ?片 ?鼎 ; 250
  93    ?克 ?彔 ?禾 ?秝 ?黍 ?香 ?米 ?毇 ?臼 ?凶 ; 260
  94    ?𣎳 ?林 ?麻 ?尗 ?耑 ?韭 ?瓜 ?瓠 ?宀 ?宮 ; 270
  95    ?呂 ?穴 ?㝱 ?𤕫 ?冖 ?𠔼 ?冃 ?㒳 ?网 ?襾 ; 280
  96    ?巾 ?巿 ?帛 ?白 ?㡀 ?黹 ?人 ?𠤎 ?匕 ?从 ; 290
  97    ?比 ?北 ?丘 ?㐺 ?𡈼 ?重 ?臥 ?身 ?㐆 ?衣 ; 300
  98    ?裘 ?老 ?毛 ?毳 ?尸 ?尺 ?尾 ?履 ?舟 ?方 ; 310
  99    ?儿 ?兄 ?兂 ?皃 ?𠑹 ?先 ?秃 ?見 ?覞 ?欠 ; 320
 100    ?㱃 ?㳄 ?旡 ?頁 ?𦣻 ?面 ?丏 ?首 ?𥄉 ?須 ; 330
 101    ?彡 ?彣 ?文 ?髟 ?后 ?司 ?卮 ?卩 ?印 ?色 ; 340
 102    ?𠨍 ?辟 ?勹 ?包 ?茍 ?鬼 ?甶 ?厶 ?嵬 ?山 ; 350
 103    ?屾 ?屵 ?广 ?厂 ?丸 ?危 ?石 ?長 ?勿 ?冄 ; 360
 104    ?而 ?豕 ?㣇 ?彑 ?豚 ?豸 ?𤉡 ?易 ?象 ?馬 ; 370
 105    ?𢊁 ?鹿 ?麤 ?㲋 ?兔 ?萈 ?犬 ?㹜 ?鼠 ?能 ; 380
 106    ?熊 ?火 ?炎 ?黑 ?囪 ?焱 ?炙 ?赤 ?大 ?亦 ; 390
 107    ])
 108
 109 (defun shuowen-radical (number)
 110   (aref shuowen-radicals (1- number)))
 111
 112 (defvar char-db-file-coding-system 'utf-8-mcs-er)
 113
 114 (defvar char-db-ignored-attributes '(ideographic-products))
 115
 116 (defvar char-db-coded-charset-priority-list
 117   '(ascii
 118     control-1
 119     latin-iso8859-1
 120     latin-iso8859-2
 121     latin-iso8859-3
 122     latin-iso8859-4
 123     latin-iso8859-9
 124     latin-jisx0201
 125     cyrillic-iso8859-5
 126     greek-iso8859-7
 127     thai-tis620
 128     ;; =mj
 129     ;; =adobe-japan1-0
 130     ;; =adobe-japan1-1
 131     ;; =adobe-japan1-2
 132     ;; =adobe-japan1-3
 133     ;; =adobe-japan1-4
 134     ;; =adobe-japan1-5
 135     ;; =adobe-japan1-6
 136     =jis-x0208
 137     =jis-x0208@1978
 138     =jis-x0208@1983
 139     japanese-jisx0212
 140     chinese-gb2312
 141     =jis-x0208@1990
 142     chinese-cns11643-1
 143     chinese-cns11643-2
 144     chinese-cns11643-3
 145     chinese-cns11643-4
 146     chinese-cns11643-5
 147     chinese-cns11643-6
 148     chinese-cns11643-7
 149     =jis-x0213-1
 150     =jis-x0213-1@2000
 151     =jis-x0213-1@2004
 152     =jis-x0213-2
 153     korean-ksc5601
 154     chinese-isoir165
 155     katakana-jisx0201
 156     hebrew-iso8859-8
 157     chinese-gb12345
 158     latin-viscii
 159     ethiopic-ucs
 160     =big5-cdp
 161     =hanyo-denshi/ja
 162     =hanyo-denshi/jb
 163     =hanyo-denshi/jc
 164     =hanyo-denshi/jd
 165     =hanyo-denshi/ft
 166     =hanyo-denshi/ia
 167     =hanyo-denshi/ib
 168     =hanyo-denshi/hg
 169     =hanyo-denshi/jt
 170     =hanyo-denshi/ks
 171     =hanyo-denshi/tk
 172     ;; ==mj
 173     ;; ==adobe-japan1-0
 174     ;; ==adobe-japan1-1
 175     ;; ==adobe-japan1-2
 176     ;; ==adobe-japan1-3
 177     ;; ==adobe-japan1-4
 178     ;; ==adobe-japan1-5
 179     ;; ==adobe-japan1-6
 180     ==jis-x0208
 181     ==jis-x0213-1
 182     ==jis-x0213-2
 183     ==jis-x0212
 184     ==hanyo-denshi/ja
 185     ==hanyo-denshi/jb
 186     ==hanyo-denshi/jc
 187     ==hanyo-denshi/ft
 188     ==hanyo-denshi/ib
 189     ==hanyo-denshi/hg
 190     ==hanyo-denshi/jt
 191     ==hanyo-denshi/ks
 192     =gt
 193     =gt-k
 194     =daikanwa
 195     =daikanwa@rev2
 196     =daikanwa@rev1
 197     =daikanwa/+p
 198     ==daikanwa
 199     =cbeta
 200     ideograph-hanziku-1
 201     ideograph-hanziku-2
 202     ideograph-hanziku-3
 203     ideograph-hanziku-4
 204     ideograph-hanziku-5
 205     ideograph-hanziku-6
 206     ideograph-hanziku-7
 207     ideograph-hanziku-8
 208     ideograph-hanziku-9
 209     ideograph-hanziku-10
 210     ideograph-hanziku-11
 211     ideograph-hanziku-12
 212     ;; =>>>adobe-japan1-0
 213     ;; =>>>adobe-japan1-1
 214     ;; =>>>adobe-japan1-2
 215     ;; =>>>adobe-japan1-3
 216     ;; =>>>adobe-japan1-4
 217     ;; =>>>adobe-japan1-5
 218     ;; =>>>adobe-japan1-6
 219     ;; =>>>jis-x0208
 220     ;; =>>>jis-x0213-1
 221     ;; =>>>jis-x0213-2
 222     ;; =>>>hanyo-denshi/ja
 223     ;; =>>>hanyo-denshi/jb
 224     ;; =>>>hanyo-denshi/jc
 225     ;; =>>>hanyo-denshi/ft
 226     ;; =>>>hanyo-denshi/ib
 227     ;; =>>>hanyo-denshi/hg
 228     ;; =>>>hanyo-denshi/jt
 229     ;; =>>>hanyo-denshi/ks
 230     ;; =>>>gt
 231     =>>adobe-japan1-0
 232     =>>adobe-japan1-1
 233     =>>adobe-japan1-2
 234     =>>adobe-japan1-3
 235     =>>adobe-japan1-4
 236     =>>adobe-japan1-5
 237     =>>adobe-japan1-6
 238     =>>jis-x0208
 239     =>>jis-x0213-1
 240     =>>jis-x0213-1@2000
 241     =>>jis-x0213-1@2004
 242     =>>jis-x0213-2
 243     =>>jis-x0208@1978
 244     =>>hanyo-denshi/ft
 245     =>>hanyo-denshi/jt
 246     =>>hanyo-denshi/ks
 247     =>>gt
 248     =>>daikanwa
 249     =>>cbeta
 250     =+>jis-x0208
 251     =+>jis-x0213-1
 252     =+>jis-x0213-2
 253     =+>adobe-japan1-0
 254     =+>adobe-japan1-1
 255     =+>adobe-japan1-2
 256     =+>adobe-japan1-3
 257     =+>adobe-japan1-4
 258     =+>adobe-japan1-5
 259     =+>adobe-japan1-6
 260     =+>jis-x0208@1978
 261     =>jis-x0208
 262     =>jis-x0208@1997
 263     =>jis-x0213-1
 264     =>jis-x0213-1@2000
 265     =>jis-x0213-1@2004
 266     =>jis-x0213-2
 267     ==>ucs@bucs
 268     =>iwds-1
 269     =>ucs@hanyo-denshi
 270     =>ucs@iso
 271     =>ucs@unicode
 272     =>ucs@jis
 273     =>ucs@cns
 274     =>ucs@ks
 275     =+>ucs@iso
 276     =+>ucs@unicode
 277     =+>ucs@jis
 278     =+>ucs@jis/1990
 279     =+>ucs@cns
 280     =+>ucs@ks
 281     =>>ucs@iso
 282     =>>ucs@unicode
 283     =>>ucs@jis
 284     =>>ucs@cns
 285     =>>>ucs@iso
 286     =>>>ucs@unicode
 287     ==ucs@iso
 288     ==ucs@unicode
 289     ;; ==ucs@cns
 290     ==gb2312
 291     ==ks-x1001
 292     ==cns11643-1
 293     ==cns11643-2
 294     ==cns11643-3
 295     ==cns11643-4
 296     ==cns11643-5
 297     ==cns11643-6
 298     ==cns11643-7
 299     ==gt
 300     ==jis-x0208@1990
 301     ;; ==jis-x0208@1983
 302     ==jis-x0208@1978
 303     ==gt-k
 304     =ucs@iso
 305     =ucs@unicode
 306     =ucs@cns
 307     ==big5-cdp
 308     ==cbeta
 309     =>>big5-cdp
 310     =>>gt-k
 311     =+>gt
 312     =>gt
 313     =>mj
 314     =>big5-cdp
 315     =>daikanwa
 316     =>daikanwa/ho
 317     =>cns11643-5
 318     =>cns11643-7
 319     =big5
 320     =big5-eten
 321     =>gt-k
 322     =zinbun-oracle
 323     =>zinbun-oracle
 324     =ruimoku-v6
 325     =>>ruimoku-v6
 326     ==ruimoku-v6
 327     =jef-china3
 328     =>cbeta
 329     =shinjigen
 330     =ucs-var-001
 331     =ucs-var-002
 332     =ucs-var-003
 333     =ucs-var-004
 334     =ucs-itaiji-001
 335     =ucs-itaiji-002
 336     =ucs-itaiji-003
 337     =ucs-itaiji-004
 338     =ucs-itaiji-005
 339     =ucs-itaiji-006
 340     =ucs-itaiji-008
 341     =big5-cdp-var-003
 342     =big5-cdp-var-004
 343     =big5-cdp-var-005
 344     =>ucs@iwds-1))
 345
 346
 347 ;;; @ char-db formatters
 348 ;;;
 349
 350 (defun char-db-make-char-spec (char)
 351   (let (ret char-spec)
 352     (cond ((characterp char)
 353            (cond ((and (setq ret (encode-char char '=ucs 'defined-only))
 354                        (not (and (<= #xE000 ret)(<= ret #xF8FF))))
 355                   (setq char-spec (list (cons '=ucs ret)))
 356                   (cond ((setq ret (get-char-attribute char 'name))
 357                          (setq char-spec (cons (cons 'name ret) char-spec))
 358                          )
 359                         ((setq ret (get-char-attribute char 'name*))
 360                          (setq char-spec (cons (cons 'name* ret) char-spec))
 361                          ))
 362                   )
 363                  ((encode-char char '=mj 'defined-only)
 364                   (setq char-spec nil)
 365                   (dolist (ccs (charset-list))
 366                     (if (and (or (eq ccs '=mj)
 367                                  ;; (eq (charset-property ccs 'iso-ir) 177)
 368                                  (string-match "=ucs@" (symbol-name ccs))
 369                                  )
 370                              (setq ccs (charset-name ccs))
 371                              (null (assq ccs char-spec))
 372                              (setq ret (encode-char char ccs 'defined-only)))
 373                         (setq char-spec (cons (cons ccs ret) char-spec))))
 374                   )
 375                  ((encode-char char '==mj 'defined-only)
 376                   (setq char-spec nil)
 377                   (dolist (ccs (charset-list))
 378                     (if (and (or (eq ccs '==mj)
 379                                  ;; (eq (charset-property ccs 'iso-ir) 177)
 380                                  (string-match "=ucs@" (symbol-name ccs))
 381                                  )
 382                              (setq ccs (charset-name ccs))
 383                              (null (assq ccs char-spec))
 384                              (setq ret (encode-char char ccs 'defined-only)))
 385                         (setq char-spec (cons (cons ccs ret) char-spec))))
 386                   )
 387                  ((encode-char char '=adobe-japan1 'defined-only)
 388                   (setq char-spec nil)
 389                   (dolist (ccs (charset-list))
 390                     (if (and (or (memq ccs
 391                                        '(=adobe-japan1-0
 392                                          =adobe-japan1-1
 393                                          =adobe-japan1-2
 394                                          =adobe-japan1-3
 395                                          =adobe-japan1-4
 396                                          =adobe-japan1-5
 397                                          =adobe-japan1-6
 398                                          ))
 399                                  ;; (eq (charset-property ccs 'iso-ir) 177)
 400                                  (string-match "=ucs@" (symbol-name ccs))
 401                                  )
 402                              (setq ccs (charset-name ccs))
 403                              (null (assq ccs char-spec))
 404                              (setq ret (encode-char char ccs 'defined-only)))
 405                         (setq char-spec (cons (cons ccs ret) char-spec))))
 406                   )
 407                  ((encode-char char '==adobe-japan1 'defined-only)
 408                   (setq char-spec nil)
 409                   (dolist (ccs (charset-list))
 410                     (if (and (or (memq ccs
 411                                        '(==adobe-japan1-0
 412                                          ==adobe-japan1-1
 413                                          ==adobe-japan1-2
 414                                          ==adobe-japan1-3
 415                                          ==adobe-japan1-4
 416                                          ==adobe-japan1-5
 417                                          ==adobe-japan1-6
 418                                          ))
 419                                  ;; (eq (charset-property ccs 'iso-ir) 177)
 420                                  (string-match "=ucs@" (symbol-name ccs))
 421                                  )
 422                              (setq ccs (charset-name ccs))
 423                              (null (assq ccs char-spec))
 424                              (setq ret (encode-char char ccs 'defined-only)))
 425                         (setq char-spec (cons (cons ccs ret) char-spec))))
 426                   )
 427                  ((setq ret
 428                         (catch 'tag
 429                           (let ((rest char-db-coded-charset-priority-list)
 430                                 ccs)
 431                             (while rest
 432                               (setq ccs (charset-name
 433                                          (find-charset (car rest))))
 434                               (if (setq ret
 435                                         (encode-char char ccs
 436                                                      'defined-only))
 437                                   (throw 'tag (cons ccs ret)))
 438                               (setq rest (cdr rest))))))
 439                   (setq char-spec (list ret))
 440                   (dolist (ccs (delq (car ret) (charset-list)))
 441                     (if (and (or (charset-iso-final-char ccs)
 442                                  (memq ccs
 443                                        '(=daikanwa
 444                                          =daikanwa@rev2
 445                                          ;; =gt-k
 446                                          =jis-x0208@1997
 447                                          ))
 448                                  (eq (charset-property ccs 'iso-ir) 177)
 449                                  ;; (string-match "=ucs@" (symbol-name ccs))
 450                                  )
 451                              (setq ccs (charset-name ccs))
 452                              (null (assq ccs char-spec))
 453                              (setq ret (encode-char char ccs 'defined-only)))
 454                         (setq char-spec (cons (cons ccs ret) char-spec))))
 455                   (if (null char-spec)
 456                       (setq char-spec (split-char char)))
 457                   (cond ((setq ret (get-char-attribute char 'name))
 458                          (setq char-spec (cons (cons 'name ret) char-spec))
 459                          )
 460                         ((setq ret (get-char-attribute char 'name*))
 461                          (setq char-spec (cons (cons 'name* ret) char-spec))
 462                          ))
 463                   )
 464                  ((setq ret (get-char-attribute
 465                              char 'ideographic-combination))
 466                   (setq char-spec
 467                         (cons (cons 'ideographic-combination ret)
 468                               char-spec))
 469                   ))
 470            char-spec)
 471           ((consp char)
 472            char))))
 473
 474 (defun char-db-insert-char-spec (char &optional readable column
 475                                       required-features)
 476   (unless column
 477     (setq column (current-column)))
 478   (let (char-spec temp-char)
 479     (setq char-spec (char-db-make-char-spec char))
 480     (unless (or (characterp char) ; char
 481                 (condition-case nil
 482                     (setq char (find-char char-spec))
 483                   (error nil)))
 484       ;; define temporary character
 485       ;;   Current implementation is dirty.
 486       (setq temp-char (define-char (cons '(ideograph-daikanwa . 0)
 487                                          char-spec)))
 488       (remove-char-attribute temp-char 'ideograph-daikanwa)
 489       (setq char temp-char))
 490     (insert-char-attributes char
 491                             readable
 492                             (union (mapcar #'car char-spec)
 493                                    required-features)
 494                             nil 'for-sub-node)
 495     (when temp-char
 496       ;; undefine temporary character
 497       ;;   Current implementation is dirty.
 498       (setq char-spec (char-attribute-alist temp-char))
 499       (while char-spec
 500         (remove-char-attribute temp-char (car (car char-spec)))
 501         (setq char-spec (cdr char-spec))))))
 502
 503 (defun char-db-insert-alist (alist &optional readable column)
 504   (unless column
 505     (setq column (current-column)))
 506   (let ((line-breaking
 507          (concat "\n" (make-string (1+ column) ?\ )))
 508         name value
 509         ret al ; cal
 510         key
 511         lbs cell rest separator)
 512     (insert "(")
 513     (while alist
 514       (setq name (car (car alist))
 515             value (cdr (car alist)))
 516       (cond ((eq name 'char)
 517              (insert "(char . ")
 518              (if (setq ret (condition-case nil
 519                                (find-char value)
 520                              (error nil)))
 521                  (progn
 522                    (setq al nil
 523                          ;; cal nil
 524                          )
 525                    (while value
 526                      (setq key (car (car value)))
 527                      ;; (if (find-charset key)
 528                      ;;     (setq cal (cons key cal))
 529                      (setq al (cons key al))
 530                      ;; )
 531                      (setq value (cdr value)))
 532                    (insert-char-attributes ret
 533                                            readable
 534                                            (or al 'none) ; cal
 535                                            nil 'for-sub-node))
 536                (insert (prin1-to-string value)))
 537              (insert ")")
 538              (insert line-breaking))
 539             ((consp value)
 540              (insert (format "(%-18s " name))
 541              (setq lbs (concat "\n" (make-string (current-column) ?\ )))
 542              (while (consp value)
 543                (setq cell (car value))
 544                (if (and (consp cell)
 545                         (consp (car cell))
 546                         (setq ret (condition-case nil
 547                                       (find-char cell)
 548                                     (error nil)))
 549                         )
 550                    (progn
 551                      (setq rest cell
 552                            al nil
 553                            ;; cal nil
 554                            )
 555                      (while rest
 556                        (setq key (car (car rest)))
 557                        ;; (if (find-charset key)
 558                        ;;     (setq cal (cons key cal))
 559                        (setq al (cons key al))
 560                        ;; )
 561                        (setq rest (cdr rest)))
 562                      (if separator
 563                          (insert lbs))
 564                      (insert-char-attributes ret
 565                                              readable
 566                                              al ; cal
 567                                              nil 'for-sub-node)
 568                      (setq separator lbs))
 569                  (if separator
 570                      (insert separator))
 571                  (insert (prin1-to-string cell))
 572                  (setq separator " "))
 573                (setq value (cdr value)))
 574              (insert ")")
 575              (insert line-breaking))
 576             (t
 577              (insert (format "(%-18s . %S)%s"
 578                              name value
 579                              line-breaking))))
 580       (setq alist (cdr alist))))
 581   (insert ")"))
 582
 583 (defun char-db-insert-char-reference (plist &optional readable column)
 584   (unless column
 585     (setq column (current-column)))
 586   (let ((line-breaking
 587          (concat "\n" (make-string (1+ column) ?\ )))
 588         (separator "")
 589         name value)
 590     (insert "(")
 591     (while plist
 592       (setq name (pop plist))
 593       (setq value (pop plist))
 594       (cond ((eq name :char)
 595              (insert separator)
 596              (insert ":char\t")
 597              (cond ((numberp value)
 598                     (setq value (decode-char '=ucs value)))
 599                    ;; ((consp value)
 600                    ;;  (setq value (or (find-char value)
 601                    ;;                  value)))
 602                    )
 603              (char-db-insert-char-spec value readable)
 604              (insert line-breaking)
 605              (setq separator ""))
 606             ((eq name :radical)
 607              (insert (format "%s%s\t%d ; %c%s"
 608                              separator
 609                              name value
 610                              (ideographic-radical value)
 611                              line-breaking))
 612              (setq separator ""))
 613             (t
 614              (insert (format "%s%s\t%S" separator name value))
 615              (setq separator line-breaking)))
 616       ))
 617   (insert ")"))
 618
 619 (defun char-db-decode-isolated-char (ccs code-point)
 620   (let (ret)
 621     (setq ret
 622           (cond ((eq ccs 'arabic-iso8859-6)
 623                  (decode-char ccs code-point))
 624                 ;; ((eq ccs '=gt)
 625                 ;;  (decode-builtin-char '==gt code-point))
 626                 ((and (memq ccs '(=gt-pj-1
 627                                   =gt-pj-2
 628                                   =gt-pj-3
 629                                   =gt-pj-4
 630                                   =gt-pj-5
 631                                   =gt-pj-6
 632                                   =gt-pj-7
 633                                   =gt-pj-8
 634                                   =gt-pj-9
 635                                   =gt-pj-10
 636                                   =gt-pj-11))
 637                       (setq ret (decode-char ccs code-point))
 638                       (setq ret (encode-char ret '=gt 'defined-only)))
 639                  (decode-builtin-char '=gt ret))
 640                 (t
 641                  (decode-builtin-char ccs code-point))))
 642     (cond ((null ret)
 643            (or (decode-char ccs code-point)
 644                (define-char (list (cons ccs code-point)))))
 645           ((and (<= 0 (char-int ret))
 646                 (<= (char-int ret) #x1F))
 647            (decode-char '=ucs (+ #x2400 (char-int ret))))
 648           ((= (char-int ret) #x7F)
 649            ?\u2421)
 650           (t ret))))
 651
 652 (defvar char-db-convert-obsolete-format t)
 653
 654 (defun char-db-insert-ccs-feature (name value line-breaking)
 655   (cond
 656    ((integerp value)
 657     (insert
 658      (format
 659       (cond
 660        ((memq name '(=>iwds-1
 661                      ==shinjigen
 662                      =shinjigen
 663                      =shinjigen@1ed ==shinjigen@1ed
 664                      =shinjigen@rev ==shinjigen@rev
 665                      =shinjigen/+p@rev ==shinjigen/+p@rev
 666                      ===daikanwa/ho ==daikanwa/ho
 667                      =daikanwa/ho =>>daikanwa/ho =>daikanwa/ho))
 668         "(%-18s .  %04d)\t; %c")
 669        ((eq name '=shinjigen@1ed/24pr)
 670         "(%-18s . %04d)\t; %c")
 671        ((or
 672          (memq name
 673                '(===daikanwa
 674                  ==daikanwa =daikanwa =>>daikanwa =>daikanwa
 675                  =daikanwa@rev1 =daikanwa@rev2
 676                  =daikanwa/+p ==daikanwa/+p ===daikanwa/+p
 677                  =>>daikanwa/+p
 678                  =daikanwa/+2p =>>daikanwa/+2p
 679                  =gt ==gt ===gt
 680                  =>>gt =+>gt =>gt
 681                  =gt-k ==gt-k ===gt-k
 682                  =>>gt-k =>gt-k
 683                  =adobe-japan1-0 ==adobe-japan1-0 ===adobe-japan1-0
 684                  =adobe-japan1-1 ==adobe-japan1-1 ===adobe-japan1-1
 685                  =adobe-japan1-2 ==adobe-japan1-2 ===adobe-japan1-2
 686                  =adobe-japan1-3 ==adobe-japan1-3 ===adobe-japan1-3
 687                  =adobe-japan1-4 ==adobe-japan1-4 ===adobe-japan1-4
 688                  =adobe-japan1-5 ==adobe-japan1-5 ===adobe-japan1-5
 689                  =adobe-japan1-6 ==adobe-japan1-6 ===adobe-japan1-6
 690                  =>>adobe-japan1-0 =+>adobe-japan1-0
 691                  =>>adobe-japan1-1 =+>adobe-japan1-1
 692                  =>>adobe-japan1-2 =+>adobe-japan1-2
 693                  =>>adobe-japan1-3 =+>adobe-japan1-3
 694                  =>>adobe-japan1-4 =+>adobe-japan1-4
 695                  =>>adobe-japan1-5 =+>adobe-japan1-5
 696                  =>>adobe-japan1-6 =+>adobe-japan1-6
 697                  =>cbeta =cbeta =>>cbeta ==cbeta ===cbeta
 698                  =zinbun-oracle =>zinbun-oracle
 699                  ===hng-jou ===hng-keg ===hng-dng ===hng-mam
 700                  ===hng-drt ===hng-kgk ===hng-myz ===hng-kda
 701                  ===hng-khi ===hng-khm ===hng-hok ===hng-kyd ===hng-sok
 702                  ===hng-yhk ===hng-kak ===hng-kar ===hng-kae
 703                  ===hng-sys ===hng-tsu ===hng-tzj
 704                  ===hng-hos ===hng-nak ===hng-jhk
 705                  ===hng-hod ===hng-gok ===hng-ink ===hng-nto
 706                  ===hng-nkm ===hng-k24 ===hng-nkk
 707                  ===hng-kcc ===hng-kcj ===hng-kbk ===hng-sik
 708                  ===hng-skk ===hng-kyu ===hng-ksk ===hng-wan
 709                  ===hng-okd ===hng-wad ===hng-kmi
 710                  ===hng-zkd ===hng-doh ===hng-jyu
 711                  ===hng-tzs ===hng-kss ===hng-kyo
 712                  ===hng-smk))
 713          ;; (string-match "^=adobe-" (symbol-name name))
 714          )
 715         "(%-18s . %05d)\t; %c")
 716        ((memq name '(=hanyo-denshi/ks
 717                      ==hanyo-denshi/ks ===hanyo-denshi/ks
 718                      =>>hanyo-denshi/ks
 719                      =koseki ==koseki
 720                      =mj ==mj ===mj =>>mj =>mj
 721                      =zihai mojikyo))
 722         "(%-18s . %06d)\t; %c")
 723        ((memq name '(=hanyo-denshi/tk ==hanyo-denshi/tk))
 724         "(%-18s . %08d)\t; %c")
 725        ((>= (charset-dimension name) 2)
 726         "(%-18s . #x%04X)\t; %c")
 727        (t
 728         "(%-18s . #x%02X)\t; %c"))
 729       name
 730       (if (= (charset-iso-graphic-plane name) 1)
 731           (logior value
 732                   (cond ((= (charset-dimension name) 1)
 733                          #x80)
 734                         ((= (charset-dimension name) 2)
 735                          #x8080)
 736                         ((= (charset-dimension name) 3)
 737                          #x808080)
 738                         (t 0)))
 739         value)
 740       (char-db-decode-isolated-char name value)))
 741     (if (and (= (charset-chars name) 94)
 742              (= (charset-dimension name) 2))
 743         (insert (format " [%02d-%02d]"
 744                         (- (lsh value -8) 32)
 745                         (- (logand value 255) 32))))
 746     )
 747    (t
 748     (insert (format "(%-18s . %s)" name value))
 749     ))
 750   (insert line-breaking))
 751
 752 (defun char-db-insert-relation-feature (char name value line-breaking
 753                                              ccss readable)
 754   (insert (format "(%-18s%s " name line-breaking))
 755   (let ((lbs (concat "\n" (make-string (current-column) ?\ )))
 756         separator cell sources required-features
 757         ret)
 758     (while (consp value)
 759       (setq cell (car value))
 760       (if (integerp cell)
 761           (setq cell (decode-char '=ucs cell)))
 762       (cond
 763        ((eq name '->subsumptive)
 764         (when (or (not (some (lambda (atr)
 765                                (get-char-attribute cell atr))
 766                              char-db-ignored-attributes))
 767                   (some (lambda (ccs)
 768                           (encode-char cell ccs 'defined-only))
 769                         ccss))
 770           (if separator
 771               (insert lbs))
 772           (let ((char-db-ignored-attributes
 773                  (cons '<-subsumptive
 774                        char-db-ignored-attributes)))
 775             (insert-char-attributes cell readable nil nil 'for-sub-node))
 776           (setq separator lbs))
 777         )
 778        ((characterp cell)
 779         (setq sources
 780               (get-char-attribute
 781                char (intern (format "%s*sources" name))))
 782         (setq required-features nil)
 783         (dolist (source sources)
 784           (cond
 785            ((memq source '(JP
 786                            JP/Jouyou
 787                            shinjigen shinjigen@1ed shinjigen@rev))
 788             (setq required-features
 789                   (union required-features
 790                          '(=jis-x0208
 791                            =jis-x0208@1990
 792                            =jis-x0213-1@2000
 793                            =jis-x0213-1@2004
 794                            =jis-x0213-2
 795                            =jis-x0212
 796                            =jis-x0208@1983
 797                            =jis-x0208@1978
 798                            =shinjigen
 799                            =shinjigen@1ed
 800                            =shinjigen@rev
 801                            =shinjigen/+p@rev))))
 802            ((eq source 'CN)
 803             (setq required-features
 804                   (union required-features
 805                          '(=gb2312
 806                            =gb12345
 807                            =iso-ir165)))))
 808           (cond
 809            ((find-charset (setq ret (intern (format "=%s" source))))
 810             (setq required-features
 811                   (cons ret required-features)))
 812            (t (setq required-features
 813                     (cons source required-features)))))
 814         (cond ((string-match "@JP" (symbol-name name))
 815                (setq required-features
 816                      (union required-features
 817                             '(=jis-x0208
 818                               =jis-x0208@1990
 819                               =jis-x0213-1-2000
 820                               =jis-x0213-2-2000
 821                               =jis-x0212
 822                               =jis-x0208@1983
 823                               =jis-x0208@1978))))
 824               ((string-match "@CN" (symbol-name name))
 825                (setq required-features
 826                      (union required-features
 827                             '(=gb2312
 828                               =gb12345
 829                               =iso-ir165)))))
 830         (if separator
 831             (insert lbs))
 832         (if readable
 833             (insert (format "%S" cell))
 834           (char-db-insert-char-spec cell readable
 835                                     nil
 836                                     required-features))
 837         (setq separator lbs))
 838        ((consp cell)
 839         (if separator
 840             (insert lbs))
 841         (if (consp (car cell))
 842             (char-db-insert-char-spec cell readable)
 843           (char-db-insert-char-reference cell readable))
 844         (setq separator lbs))
 845        (t
 846         (if separator
 847             (insert separator))
 848         (insert (prin1-to-string cell))
 849         (setq separator " ")))
 850       (setq value (cdr value)))
 851     (insert ")")
 852     (insert line-breaking)))
 853
 854 (defun insert-char-attributes (char &optional readable attributes column
 855                                     for-sub-node)
 856   (unless column
 857     (setq column (current-column)))
 858   (let (name value ; has-long-ccs-name
 859         rest
 860         radical strokes
 861         (line-breaking
 862          (concat "\n" (make-string (1+ column) ?\ )))
 863         lbs cell separator ret
 864         key al cal
 865         dest-ccss ; sources required-features
 866         ccss)
 867     (let (atr-d)
 868       (setq attributes
 869             (sort (if attributes
 870                       (if (consp attributes)
 871                           (progn
 872                             (dolist (name attributes)
 873                               (unless (memq name char-db-ignored-attributes)
 874                                 (if (find-charset name)
 875                                     (push name ccss))
 876                                 (push name atr-d)))
 877                             atr-d))
 878                     (dolist (name (char-attribute-list))
 879                       (unless (memq name char-db-ignored-attributes)
 880                         (if (find-charset name)
 881                             (push name ccss))
 882                         (push name atr-d)))
 883                     atr-d)
 884                   #'char-attribute-name<)))
 885     (insert "(")
 886     (when (memq '<-subsumptive attributes)
 887       (when (or readable (not for-sub-node))
 888         (when (setq value (get-char-attribute char '<-subsumptive))
 889           (char-db-insert-relation-feature char '<-subsumptive value
 890                                            line-breaking
 891                                            ccss readable)))
 892       (setq attributes (delq '<-subsumptive attributes)))
 893     (when (and (memq '<-denotational attributes)
 894                (setq value (get-char-attribute char '<-denotational)))
 895       (char-db-insert-relation-feature char '<-denotational value
 896                                        line-breaking
 897                                        ccss readable)
 898       (setq attributes (delq '<-denotational attributes)))
 899     (when (and (memq '<-denotational@component attributes)
 900                (setq value (get-char-attribute char '<-denotational@component)))
 901       (char-db-insert-relation-feature char '<-denotational@component value
 902                                        line-breaking
 903                                        ccss readable)
 904       (setq attributes (delq '<-denotational@component attributes)))
 905     (when (and (memq 'name attributes)
 906                (setq value (get-char-attribute char 'name)))
 907       (insert (format
 908                (if (> (+ (current-column) (length value)) 48)
 909                    "(name . %S)%s"
 910                  "(name               . %S)%s")
 911                value line-breaking))
 912       (setq attributes (delq 'name attributes))
 913       )
 914     (when (and (memq 'name* attributes)
 915                (setq value (get-char-attribute char 'name*)))
 916       (insert (format
 917                (if (> (+ (current-column) (length value)) 48)
 918                    "(name* . %S)%s"
 919                  "(name*              . %S)%s")
 920                value line-breaking))
 921       (setq attributes (delq 'name* attributes))
 922       )
 923     (when (and (memq 'script attributes)
 924                (setq value (get-char-attribute char 'script)))
 925       (insert (format "(script\t\t%s)%s"
 926                       (mapconcat (function prin1-to-string)
 927                                  value " ")
 928                       line-breaking))
 929       (setq attributes (delq 'script attributes))
 930       )
 931     (dolist (name '(=>ucs =>ucs*))
 932       (when (and (memq name attributes)
 933                  (setq value (get-char-attribute char name)))
 934         (insert (format "(%-18s . #x%04X)\t; %c%s"
 935                         name value (decode-char '=ucs value)
 936                         line-breaking))
 937         (setq attributes (delq name attributes))))
 938     (dolist (name '(=>ucs@gb =>ucs@big5))
 939       (when (and (memq name attributes)
 940                  (setq value (get-char-attribute char name)))
 941         (insert (format "(%-18s . #x%04X)\t; %c%s"
 942                         name value
 943                         (decode-char (intern
 944                                       (concat "="
 945                                               (substring
 946                                                (symbol-name name) 2)))
 947                                      value)
 948                         line-breaking))
 949         (setq attributes (delq name attributes))
 950         ))
 951     ;; (dolist (name '(=>daikanwa))
 952     ;;   (when (and (memq name attributes)
 953     ;;              (setq value (get-char-attribute char name)))
 954     ;;     (insert
 955     ;;      (if (integerp value)
 956     ;;          (format "(%-18s . %05d)\t; %c%s"
 957     ;;                  name value (decode-char '=daikanwa value)
 958     ;;                  line-breaking)
 959     ;;        (format "(%-18s %s)\t; %c%s"
 960     ;;                name
 961     ;;                (mapconcat (function prin1-to-string)
 962     ;;                           value " ")
 963     ;;                (char-representative-of-daikanwa char)
 964     ;;                line-breaking)))
 965     ;;     (setq attributes (delq name attributes))))
 966     (when (and (memq 'general-category attributes)
 967                (setq value (get-char-attribute char 'general-category)))
 968       (insert (format
 969                "(general-category\t%s) ; %s%s"
 970                (mapconcat (lambda (cell)
 971                             (format "%S" cell))
 972                           value " ")
 973                (cond ((rassoc value unidata-normative-category-alist)
 974                       "Normative Category")
 975                      ((rassoc value unidata-informative-category-alist)
 976                       "Informative Category")
 977                      (t
 978                       "Unknown Category"))
 979                line-breaking))
 980       (setq attributes (delq 'general-category attributes))
 981       )
 982     (when (and (memq 'bidi-category attributes)
 983                (setq value (get-char-attribute char 'bidi-category)))
 984       (insert (format "(bidi-category\t. %S)%s"
 985                       value
 986                       line-breaking))
 987       (setq attributes (delq 'bidi-category attributes))
 988       )
 989     (unless (or (not (memq 'mirrored attributes))
 990                 (eq (setq value (get-char-attribute char 'mirrored 'empty))
 991                     'empty))
 992       (insert (format "(mirrored\t\t. %S)%s"
 993                       value
 994                       line-breaking))
 995       (setq attributes (delq 'mirrored attributes))
 996       )
 997     (cond
 998      ((and (memq 'decimal-digit-value attributes)
 999            (setq value (get-char-attribute char 'decimal-digit-value)))
1000       (insert (format "(decimal-digit-value . %S)%s"
1001                       value
1002                       line-breaking))
1003       (setq attributes (delq 'decimal-digit-value attributes))
1004       (when (and (memq 'digit-value attributes)
1005                  (setq value (get-char-attribute char 'digit-value)))
1006         (insert (format "(digit-value\t . %S)%s"
1007                         value
1008                         line-breaking))
1009         (setq attributes (delq 'digit-value attributes))
1010         )
1011       (when (and (memq 'numeric-value attributes)
1012                  (setq value (get-char-attribute char 'numeric-value)))
1013         (insert (format "(numeric-value\t . %S)%s"
1014                         value
1015                         line-breaking))
1016         (setq attributes (delq 'numeric-value attributes))
1017         )
1018       )
1019      (t
1020       (when (and (memq 'digit-value attributes)
1021                  (setq value (get-char-attribute char 'digit-value)))
1022         (insert (format "(digit-value\t. %S)%s"
1023                         value
1024                         line-breaking))
1025         (setq attributes (delq 'digit-value attributes))
1026         )
1027       (when (and (memq 'numeric-value attributes)
1028                  (setq value (get-char-attribute char 'numeric-value)))
1029         (insert (format "(numeric-value\t. %S)%s"
1030                         value
1031                         line-breaking))
1032         (setq attributes (delq 'numeric-value attributes))
1033         )))
1034     (when (and (memq 'iso-10646-comment attributes)
1035                (setq value (get-char-attribute char 'iso-10646-comment)))
1036       (insert (format "(iso-10646-comment\t. %S)%s"
1037                       value
1038                       line-breaking))
1039       (setq attributes (delq 'iso-10646-comment attributes))
1040       )
1041     (when (and (memq 'morohashi-daikanwa attributes)
1042                (setq value (get-char-attribute char 'morohashi-daikanwa)))
1043       (insert (format "(morohashi-daikanwa\t%s)%s"
1044                       (mapconcat (function prin1-to-string) value " ")
1045                       line-breaking))
1046       (setq attributes (delq 'morohashi-daikanwa attributes))
1047       )
1048     (setq radical nil
1049           strokes nil)
1050     (when (and (memq 'ideographic-radical attributes)
1051                (setq value (get-char-attribute char 'ideographic-radical)))
1052       (setq radical value)
1053       (insert (format "(ideographic-radical . %S)\t; %c%s"
1054                       radical
1055                       (ideographic-radical radical)
1056                       line-breaking))
1057       (setq attributes (delq 'ideographic-radical attributes))
1058       )
1059     (when (and (memq 'shuowen-radical attributes)
1060                (setq value (get-char-attribute char 'shuowen-radical)))
1061       (insert (format "(shuowen-radical\t. %S)\t; %c%s"
1062                       value
1063                       (shuowen-radical value)
1064                       line-breaking))
1065       (setq attributes (delq 'shuowen-radical attributes))
1066       )
1067     (let (key)
1068       (dolist (domain
1069                (append
1070                 char-db-feature-domains
1071                 (let (dest domain)
1072                   (dolist (feature (char-attribute-list))
1073                     (setq feature (symbol-name feature))
1074                     (when (string-match
1075                            "\\(radical\\|strokes\\)@\\([^@*]+\\)\\(\\*\\|$\\)"
1076                            feature)
1077                       (setq domain (intern (match-string 2 feature)))
1078                      (unless (memq domain dest)
1079                        (setq dest (cons domain dest)))))
1080                   (sort dest #'string<))))
1081         (setq key (intern (format "%s@%s" 'ideographic-radical domain)))
1082         (when (and (memq key attributes)
1083                    (setq value (get-char-attribute char key)))
1084           (setq radical value)
1085           (insert (format "(%s . %S)\t; %c%s"
1086                           key
1087                           radical
1088                           (ideographic-radical radical)
1089                           line-breaking))
1090           (setq attributes (delq key attributes))
1091           )
1092         (setq key (intern (format "%s@%s" 'ideographic-strokes domain)))
1093         (when (and (memq key attributes)
1094                    (setq value (get-char-attribute char key)))
1095           (setq strokes value)
1096           (insert (format "(%s . %S)%s"
1097                           key
1098                           strokes
1099                           line-breaking))
1100           (setq attributes (delq key attributes))
1101           )
1102         (setq key (intern (format "%s@%s" 'total-strokes domain)))
1103         (when (and (memq key attributes)
1104                    (setq value (get-char-attribute char key)))
1105           (insert (format "(%s       . %S)%s"
1106                           key
1107                           value
1108                           line-breaking))
1109           (setq attributes (delq key attributes))
1110           )
1111         (dolist (feature '(ideographic-radical
1112                            ideographic-strokes
1113                            total-strokes))
1114           (setq key (intern (format "%s@%s*sources" feature domain)))
1115           (when (and (memq key attributes)
1116                      (setq value (get-char-attribute char key)))
1117             (insert (format "(%s%s" key line-breaking))
1118             (dolist (cell value)
1119               (insert (format " %s" cell)))
1120             (insert ")")
1121             (insert line-breaking)
1122             (setq attributes (delq key attributes))
1123             ))
1124         ))
1125     (when (and (memq 'ideographic-strokes attributes)
1126                (setq value (get-char-attribute char 'ideographic-strokes)))
1127       (setq strokes value)
1128       (insert (format "(ideographic-strokes . %S)%s"
1129                       strokes
1130                       line-breaking))
1131       (setq attributes (delq 'ideographic-strokes attributes))
1132       )
1133     (when (and (memq 'kangxi-radical attributes)
1134                (setq value (get-char-attribute char 'kangxi-radical)))
1135       (unless (eq value radical)
1136         (insert (format "(kangxi-radical\t . %S)\t; %c%s"
1137                         value
1138                         (ideographic-radical value)
1139                         line-breaking))
1140         (or radical
1141             (setq radical value)))
1142       (setq attributes (delq 'kangxi-radical attributes))
1143       )
1144     (when (and (memq 'kangxi-strokes attributes)
1145                (setq value (get-char-attribute char 'kangxi-strokes)))
1146       (unless (eq value strokes)
1147         (insert (format "(kangxi-strokes\t . %S)%s"
1148                         value
1149                         line-breaking))
1150         (or strokes
1151             (setq strokes value)))
1152       (setq attributes (delq 'kangxi-strokes attributes))
1153       )
1154     (when (and (memq 'japanese-radical attributes)
1155                (setq value (get-char-attribute char 'japanese-radical)))
1156       (unless (eq value radical)
1157         (insert (format "(japanese-radical\t . %S)\t; %c%s"
1158                         value
1159                         (ideographic-radical value)
1160                         line-breaking))
1161         (or radical
1162             (setq radical value)))
1163       (setq attributes (delq 'japanese-radical attributes))
1164       )
1165     (when (and (memq 'japanese-strokes attributes)
1166                (setq value (get-char-attribute char 'japanese-strokes)))
1167       (unless (eq value strokes)
1168         (insert (format "(japanese-strokes\t . %S)%s"
1169                         value
1170                         line-breaking))
1171         (or strokes
1172             (setq strokes value)))
1173       (setq attributes (delq 'japanese-strokes attributes))
1174       )
1175     (when (and (memq 'cns-radical attributes)
1176                (setq value (get-char-attribute char 'cns-radical)))
1177       (insert (format "(cns-radical\t . %S)\t; %c%s"
1178                       value
1179                       (ideographic-radical value)
1180                       line-breaking))
1181       (setq attributes (delq 'cns-radical attributes))
1182       )
1183     (when (and (memq 'cns-strokes attributes)
1184                (setq value (get-char-attribute char 'cns-strokes)))
1185       (unless (eq value strokes)
1186         (insert (format "(cns-strokes\t . %S)%s"
1187                         value
1188                         line-breaking))
1189         (or strokes
1190             (setq strokes value)))
1191       (setq attributes (delq 'cns-strokes attributes))
1192       )
1193     ;; (when (and (memq 'shinjigen-1-radical attributes)
1194     ;;            (setq value (get-char-attribute char 'shinjigen-1-radical)))
1195     ;;   (unless (eq value radical)
1196     ;;     (insert (format "(shinjigen-1-radical . %S)\t; %c%s"
1197     ;;                     value
1198     ;;                     (ideographic-radical value)
1199     ;;                     line-breaking))
1200     ;;     (or radical
1201     ;;         (setq radical value)))
1202     ;;   (setq attributes (delq 'shinjigen-1-radical attributes))
1203     ;;   )
1204     (when (and (memq 'ideographic- attributes)
1205                (setq value (get-char-attribute char 'ideographic-)))
1206       (insert "(ideographic-       ")
1207       (setq lbs (concat "\n" (make-string (current-column) ?\ ))
1208             separator nil)
1209       (while (consp value)
1210         (setq cell (car value))
1211         (if (integerp cell)
1212             (setq cell (decode-char '=ucs cell)))
1213         (cond ((characterp cell)
1214                (if separator
1215                    (insert lbs))
1216                (if readable
1217                    (insert (format "%S" cell))
1218                  (char-db-insert-char-spec cell readable))
1219                (setq separator lbs))
1220               ((consp cell)
1221                (if separator
1222                    (insert lbs))
1223                (if (consp (car cell))
1224                    (char-db-insert-char-spec cell readable)
1225                  (char-db-insert-char-reference cell readable))
1226                (setq separator lbs))
1227               (t
1228                (if separator
1229                    (insert separator))
1230                (insert (prin1-to-string cell))
1231                (setq separator " ")))
1232         (setq value (cdr value)))
1233       (insert ")")
1234       (insert line-breaking)
1235       (setq attributes (delq 'ideographic- attributes)))
1236     (when (and (memq 'total-strokes attributes)
1237                (setq value (get-char-attribute char 'total-strokes)))
1238       (insert (format "(total-strokes       . %S)%s"
1239                       value
1240                       line-breaking))
1241       (setq attributes (delq 'total-strokes attributes))
1242       )
1243     (when (and (memq '->ideograph attributes)
1244                (setq value (get-char-attribute char '->ideograph)))
1245       (insert (format "(->ideograph\t%s)%s"
1246                       (mapconcat (lambda (code)
1247                                    (cond ((symbolp code)
1248                                           (symbol-name code))
1249                                          ((integerp code)
1250                                           (format "#x%04X" code))
1251                                          (t
1252                                           (format "%s %S"
1253                                                   line-breaking code))))
1254                                  value " ")
1255                       line-breaking))
1256       (setq attributes (delq '->ideograph attributes))
1257       )
1258     ;; (when (and (memq '->decomposition attributes)
1259     ;;            (setq value (get-char-attribute char '->decomposition)))
1260     ;;   (insert (format "(->decomposition\t%s)%s"
1261     ;;                   (mapconcat (lambda (code)
1262     ;;                                (cond ((symbolp code)
1263     ;;                                       (symbol-name code))
1264     ;;                                      ((characterp code)
1265     ;;                                       (if readable
1266     ;;                                           (format "%S" code)
1267     ;;                                         (format "#x%04X"
1268     ;;                                                 (char-int code))
1269     ;;                                         ))
1270     ;;                                      ((integerp code)
1271     ;;                                       (format "#x%04X" code))
1272     ;;                                      (t
1273     ;;                                       (format "%s%S" line-breaking code))))
1274     ;;                              value " ")
1275     ;;                   line-breaking))
1276     ;;   (setq attributes (delq '->decomposition attributes))
1277     ;;   )
1278     (if (equal (get-char-attribute char '->titlecase)
1279                (get-char-attribute char '->uppercase))
1280         (setq attributes (delq '->titlecase attributes)))
1281     (when (and (memq '->mojikyo attributes)
1282                (setq value (get-char-attribute char '->mojikyo)))
1283       (insert (format "(->mojikyo\t\t. %06d)\t; %c%s"
1284                       value (decode-char 'mojikyo value)
1285                       line-breaking))
1286       (setq attributes (delq '->mojikyo attributes))
1287       )
1288     (when (and (memq 'hanyu-dazidian-vol attributes)
1289                (setq value (get-char-attribute char 'hanyu-dazidian-vol)))
1290       (insert (format "(hanyu-dazidian-vol  . %d)%s"
1291                       value line-breaking))
1292       (setq attributes (delq 'hanyu-dazidian-vol attributes))
1293       )
1294     (when (and (memq 'hanyu-dazidian-page attributes)
1295                (setq value (get-char-attribute char 'hanyu-dazidian-page)))
1296       (insert (format "(hanyu-dazidian-page . %d)%s"
1297                       value line-breaking))
1298       (setq attributes (delq 'hanyu-dazidian-page attributes))
1299       )
1300     (when (and (memq 'hanyu-dazidian-char attributes)
1301                (setq value (get-char-attribute char 'hanyu-dazidian-char)))
1302       (insert (format "(hanyu-dazidian-char . %d)%s"
1303                       value line-breaking))
1304       (setq attributes (delq 'hanyu-dazidian-char attributes))
1305       )
1306     (unless readable
1307       (dolist (ignored '(composition
1308                          ->denotational <-subsumptive ->ucs-unified
1309                          ->ideographic-component-forms))
1310         (setq attributes (delq ignored attributes))))
1311     (while attributes
1312       (setq name (car attributes))
1313       (unless (eq (setq value (get-char-attribute char name 'value-is-empty))
1314                   'value-is-empty)
1315         (cond ((setq ret (find-charset name))
1316                (setq name (charset-name ret))
1317                (when (not (memq name dest-ccss))
1318                  (setq dest-ccss (cons name dest-ccss))
1319                  (char-db-insert-ccs-feature name value line-breaking))
1320                )
1321               ((string-match "^=>ucs@" (symbol-name name))
1322                (insert (format "(%-18s . #x%04X)\t; %c%s"
1323                                name value (decode-char '=ucs value)
1324                                line-breaking))
1325                )
1326               ((eq name 'jisx0208-1978/4X)
1327                (insert (format "(%-18s . #x%04X)%s"
1328                                name value
1329                                line-breaking))
1330                )
1331               ((and
1332                 (not readable)
1333                 (not (eq name '->subsumptive))
1334                 (not (eq name '->uppercase))
1335                 (not (eq name '->lowercase))
1336                 (not (eq name '->titlecase))
1337                 (not (eq name '->canonical))
1338                 (not (eq name '->Bopomofo))
1339                 (not (eq name '->mistakable))
1340                 (not (eq name '->ideographic-variants))
1341                 (null (get-char-attribute
1342                        char (intern (format "%s*sources" name))))
1343                 (not (string-match "\\*sources$" (symbol-name name)))
1344                 (null (get-char-attribute
1345                        char (intern (format "%s*note" name))))
1346                 (not (string-match "\\*note$" (symbol-name name)))
1347                 (or (eq name '<-identical)
1348                     (eq name '<-uppercase)
1349                     (eq name '<-lowercase)
1350                     (eq name '<-titlecase)
1351                     (eq name '<-canonical)
1352                     (eq name '<-ideographic-variants)
1353                     ;; (eq name '<-synonyms)
1354                     (string-match "^<-synonyms" (symbol-name name))
1355                     (eq name '<-mistakable)
1356                     (when (string-match "^->" (symbol-name name))
1357                       (cond
1358                        ((string-match "^->fullwidth" (symbol-name name))
1359                         (not (and (consp value)
1360                                   (characterp (car value))
1361                                   (encode-char
1362                                    (car value) '=ucs 'defined-only)))
1363                         )
1364                        (t)))
1365                     ))
1366                )
1367               ((or (eq name 'ideographic-structure)
1368                    (eq name 'ideographic-combination)
1369                    (eq name 'ideographic-)
1370                    (eq name '=decomposition)
1371                    (char-feature-base-name= '=decomposition name)
1372                    (char-feature-base-name= '=>decomposition name)
1373                    ;; (string-match "^=>*decomposition\\(@[^*]+\\)?$"
1374                    ;;               (symbol-name name))
1375                    (string-match "^\\(->\\|<-\\)[^*]*$" (symbol-name name))
1376                    (string-match "^\\(->\\|<-\\)[^*]*\\*sources$"
1377                                  (symbol-name name))
1378                    )
1379                (char-db-insert-relation-feature char name value
1380                                                 line-breaking
1381                                                 ccss readable))
1382               ((memq name '(ideograph=
1383                             original-ideograph-of
1384                             ancient-ideograph-of
1385                             vulgar-ideograph-of
1386                             wrong-ideograph-of
1387                             ;; simplified-ideograph-of
1388                             ideographic-variants
1389                             ;; ideographic-different-form-of
1390                             ))
1391                (insert (format "(%-18s%s " name line-breaking))
1392                (setq lbs (concat "\n" (make-string (current-column) ?\ ))
1393                      separator nil)
1394                (while (consp value)
1395                  (setq cell (car value))
1396                  (if (and (consp cell)
1397                           (consp (car cell)))
1398                      (progn
1399                        (if separator
1400                            (insert lbs))
1401                        (char-db-insert-alist cell readable)
1402                        (setq separator lbs))
1403                    (if separator
1404                        (insert separator))
1405                    (insert (prin1-to-string cell))
1406                    (setq separator " "))
1407                  (setq value (cdr value)))
1408                (insert ")")
1409                (insert line-breaking))
1410               ((consp value)
1411                (insert (format "(%-18s " name))
1412                (setq lbs (concat "\n" (make-string (current-column) ?\ ))
1413                      separator nil)
1414                (while (consp value)
1415                  (setq cell (car value))
1416                  (if (and (consp cell)
1417                           (consp (car cell))
1418                           (setq ret (condition-case nil
1419                                         (find-char cell)
1420                                       (error nil))))
1421                      (progn
1422                        (setq rest cell
1423                              al nil
1424                              cal nil)
1425                        (while rest
1426                          (setq key (car (car rest)))
1427                          (if (find-charset key)
1428                              (setq cal (cons key cal))
1429                            (setq al (cons key al)))
1430                          (setq rest (cdr rest)))
1431                        (if separator
1432                            (insert lbs))
1433                        (insert-char-attributes ret
1434                                                readable
1435                                                al ; cal
1436                                                nil 'for-sub-node)
1437                        (setq separator lbs))
1438                    (setq ret (prin1-to-string cell))
1439                    (if separator
1440                        (if (< (+ (current-column)
1441                                  (length ret)
1442                                  (length separator))
1443                               76)
1444                            (insert separator)
1445                          (insert lbs)))
1446                    (insert ret)
1447                    (setq separator " "))
1448                  (setq value (cdr value)))
1449                (insert ")")
1450                (insert line-breaking))
1451               (t
1452                (insert (format "(%-18s" name))
1453                (setq ret (prin1-to-string value))
1454                (unless (< (+ (current-column)
1455                              (length ret)
1456                              3)
1457                           76)
1458                  (insert line-breaking))
1459                (insert " . " ret ")" line-breaking)
1460                ;; (insert (format "(%-18s . %S)%s"
1461                ;;                 name value
1462                ;;                 line-breaking))
1463                )
1464               ))
1465       (setq attributes (cdr attributes)))
1466     (insert ")")))
1467
1468 (defun insert-char-data (char &optional readable
1469                               attributes)
1470   (save-restriction
1471     (narrow-to-region (point)(point))
1472     (insert "(define-char
1473   '")
1474     (insert-char-attributes char readable attributes)
1475     (insert ")\n")
1476     (goto-char (point-min))
1477     (while (re-search-forward "[ \t]+$" nil t)
1478       (replace-match ""))
1479     ;; from tabify.
1480     (goto-char (point-min))
1481     (while (re-search-forward "[ \t][ \t][ \t]*" nil t)
1482       (let ((column (current-column))
1483             (indent-tabs-mode t))
1484         (delete-region (match-beginning 0) (point))
1485         (indent-to column)))
1486     (goto-char (point-max))
1487     ;; (tabify (point-min)(point-max))
1488     ))
1489
1490 (defun insert-char-data-with-variant (char &optional printable
1491                                            no-ucs-unified
1492                                            script excluded-script)
1493   (insert-char-data char printable)
1494   (let ((variants (char-variants char))
1495         rest
1496         variant vs ret)
1497     (setq variants (sort variants #'<))
1498     (setq rest variants)
1499     (setq variants (cons char variants))
1500     (while rest
1501       (setq variant (car rest))
1502       (unless (get-char-attribute variant '<-subsumptive)
1503         (if (and (or (null script)
1504                      (null (setq vs (get-char-attribute variant 'script)))
1505                      (memq script vs))
1506                  (or (null excluded-script)
1507                      (null (setq vs (get-char-attribute variant 'script)))
1508                      (not (memq excluded-script vs))))
1509             (unless (and no-ucs-unified (get-char-attribute variant '=ucs))
1510               (insert-char-data variant printable)
1511               (if (setq ret (char-variants variant))
1512                   (while ret
1513                     (or (memq (car ret) variants)
1514                         ;; (get-char-attribute (car ret) '<-subsumptive)
1515                         (setq rest (nconc rest (list (car ret)))))
1516                     (setq ret (cdr ret)))))))
1517       (setq rest (cdr rest)))))
1518
1519 (defun insert-char-range-data (min max &optional script excluded-script)
1520   (let ((code min)
1521         char)
1522     (while (<= code max)
1523       (setq char (decode-char '=ucs code))
1524       (if (encode-char char '=ucs 'defined-only)
1525           (insert-char-data-with-variant char nil 'no-ucs-unified
1526                                          script excluded-script))
1527       (setq code (1+ code)))))
1528
1529 (defun write-char-range-data-to-file (min max file
1530                                           &optional script excluded-script)
1531   (let ((coding-system-for-write char-db-file-coding-system))
1532     (with-temp-buffer
1533       (insert (format ";; -*- coding: %s -*-\n"
1534                       char-db-file-coding-system))
1535       (insert-char-range-data min max script excluded-script)
1536       (write-region (point-min)(point-max) file))))
1537
1538 (defvar what-character-original-window-configuration)
1539
1540 ;;;###autoload
1541 (defun what-char-definition (char)
1542   (interactive (list (char-after)))
1543   (let ((buf (get-buffer-create "*Character Description*"))
1544         (the-buf (current-buffer))
1545         (win-conf (current-window-configuration)))
1546     (pop-to-buffer buf)
1547     (make-local-variable 'what-character-original-window-configuration)
1548     (setq what-character-original-window-configuration win-conf)
1549     (setq buffer-read-only nil)
1550     (erase-buffer)
1551     (condition-case err
1552         (progn
1553           (insert-char-data-with-variant char 'printable)
1554           (unless (char-attribute-alist char)
1555             (insert (format ";; = %c\n"
1556                             (let* ((rest (split-char char))
1557                                    (ccs (pop rest))
1558                                    (code (pop rest)))
1559                               (while rest
1560                                 (setq code (logior (lsh code 8)
1561                                                    (pop rest))))
1562                               (decode-char ccs code)))))
1563           ;; (char-db-update-comment)
1564           (set-buffer-modified-p nil)
1565           (view-mode the-buf (lambda (buf)
1566                                (set-window-configuration
1567                                 what-character-original-window-configuration)
1568                                ))
1569           (goto-char (point-min)))
1570       (error (progn
1571                (set-window-configuration
1572                 what-character-original-window-configuration)
1573                (signal (car err) (cdr err)))))))
1574
1575
1576 ;;; @ end
1577 ;;;
1578
1579 (provide 'char-db-util)
1580
1581 ;;; char-db-util.el ends here