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

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