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,
   4 ;;   2007, 2008, 2009, 2010, 2011, 2012, 2013, 2014 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    ])
  99
 100 (defun shuowen-radical (number)
 101   (aref shuowen-radicals (1- number)))
 102
 103 (defvar char-db-file-coding-system 'utf-8-mcs-er)
 104
 105 (defvar char-db-ignored-attributes '(ideographic-products))
 106
 107 (defvar char-db-coded-charset-priority-list
 108   '(ascii
 109     control-1
 110     latin-iso8859-1
 111     latin-iso8859-2
 112     latin-iso8859-3
 113     latin-iso8859-4
 114     latin-iso8859-9
 115     latin-jisx0201
 116     cyrillic-iso8859-5
 117     greek-iso8859-7
 118     thai-tis620
 119     ;; =mj
 120     ;; =adobe-japan1-0
 121     ;; =adobe-japan1-1
 122     ;; =adobe-japan1-2
 123     ;; =adobe-japan1-3
 124     ;; =adobe-japan1-4
 125     ;; =adobe-japan1-5
 126     ;; =adobe-japan1-6
 127     =jis-x0208
 128     =jis-x0208@1978
 129     =jis-x0208@1983
 130     japanese-jisx0212
 131     chinese-gb2312
 132     =jis-x0208@1990
 133     chinese-cns11643-1
 134     chinese-cns11643-2
 135     chinese-cns11643-3
 136     chinese-cns11643-4
 137     chinese-cns11643-5
 138     chinese-cns11643-6
 139     chinese-cns11643-7
 140     =jis-x0213-1
 141     =jis-x0213-1@2000
 142     =jis-x0213-1@2004
 143     =jis-x0213-2
 144     korean-ksc5601
 145     chinese-isoir165
 146     katakana-jisx0201
 147     hebrew-iso8859-8
 148     chinese-gb12345
 149     latin-viscii
 150     ethiopic-ucs
 151     =big5-cdp
 152     =hanyo-denshi/ja
 153     =hanyo-denshi/jb
 154     =hanyo-denshi/jc
 155     =hanyo-denshi/jd
 156     =hanyo-denshi/ft
 157     =hanyo-denshi/ia
 158     =hanyo-denshi/ib
 159     =hanyo-denshi/hg
 160     =hanyo-denshi/jt
 161     =hanyo-denshi/ks
 162     ;; ==mj
 163     ;; ==adobe-japan1-0
 164     ;; ==adobe-japan1-1
 165     ;; ==adobe-japan1-2
 166     ;; ==adobe-japan1-3
 167     ;; ==adobe-japan1-4
 168     ;; ==adobe-japan1-5
 169     ;; ==adobe-japan1-6
 170     ==jis-x0208
 171     ==jis-x0213-1
 172     ==jis-x0213-2
 173     ==hanyo-denshi/ja
 174     ==hanyo-denshi/jb
 175     ==hanyo-denshi/jc
 176     ==hanyo-denshi/ft
 177     ==hanyo-denshi/ib
 178     ==hanyo-denshi/hg
 179     ==hanyo-denshi/jt
 180     ==hanyo-denshi/ks
 181     =gt
 182     =gt-k
 183     =daikanwa
 184     =daikanwa@rev2
 185     =daikanwa@rev1
 186     ==daikanwa
 187     =cbeta
 188     ideograph-hanziku-1
 189     ideograph-hanziku-2
 190     ideograph-hanziku-3
 191     ideograph-hanziku-4
 192     ideograph-hanziku-5
 193     ideograph-hanziku-6
 194     ideograph-hanziku-7
 195     ideograph-hanziku-8
 196     ideograph-hanziku-9
 197     ideograph-hanziku-10
 198     ideograph-hanziku-11
 199     ideograph-hanziku-12
 200     ;; =>>>adobe-japan1-0
 201     ;; =>>>adobe-japan1-1
 202     ;; =>>>adobe-japan1-2
 203     ;; =>>>adobe-japan1-3
 204     ;; =>>>adobe-japan1-4
 205     ;; =>>>adobe-japan1-5
 206     ;; =>>>adobe-japan1-6
 207     ;; =>>>jis-x0208
 208     ;; =>>>jis-x0213-1
 209     ;; =>>>jis-x0213-2
 210     ;; =>>>hanyo-denshi/ja
 211     ;; =>>>hanyo-denshi/jb
 212     ;; =>>>hanyo-denshi/jc
 213     ;; =>>>hanyo-denshi/ft
 214     ;; =>>>hanyo-denshi/ib
 215     ;; =>>>hanyo-denshi/hg
 216     ;; =>>>hanyo-denshi/jt
 217     ;; =>>>hanyo-denshi/ks
 218     ;; =>>>gt
 219     =>>adobe-japan1-0
 220     =>>adobe-japan1-1
 221     =>>adobe-japan1-2
 222     =>>adobe-japan1-3
 223     =>>adobe-japan1-4
 224     =>>adobe-japan1-5
 225     =>>adobe-japan1-6
 226     =>>jis-x0208
 227     =>>jis-x0213-1
 228     =>>jis-x0213-1@2000
 229     =>>jis-x0213-1@2004
 230     =>>jis-x0213-2
 231     =>>jis-x0208@1978
 232     =>>hanyo-denshi/ft
 233     =>>hanyo-denshi/jt
 234     =>>hanyo-denshi/ks
 235     =>>gt
 236     =>>daikanwa
 237     =>>cbeta
 238     =+>jis-x0208
 239     =+>jis-x0213-1
 240     =+>jis-x0213-2
 241     =+>adobe-japan1-0
 242     =+>adobe-japan1-1
 243     =+>adobe-japan1-2
 244     =+>adobe-japan1-3
 245     =+>adobe-japan1-4
 246     =+>adobe-japan1-5
 247     =+>adobe-japan1-6
 248     =+>jis-x0208@1978
 249     =>jis-x0208
 250     =>jis-x0208@1997
 251     =>jis-x0213-1
 252     =>jis-x0213-1@2000
 253     =>jis-x0213-1@2004
 254     =>jis-x0213-2
 255     ==>ucs@bucs
 256     =>ucs@hanyo-denshi
 257     =>ucs@iso
 258     =>ucs@unicode
 259     =>ucs@jis
 260     =>ucs@cns
 261     =>ucs@ks
 262     =+>ucs@iso
 263     =+>ucs@unicode
 264     =+>ucs@jis
 265     =+>ucs@jis/1990
 266     =+>ucs@cns
 267     =+>ucs@ks
 268     =>>ucs@iso
 269     =>>ucs@unicode
 270     =>>ucs@jis
 271     =>>ucs@cns
 272     =>>>ucs@iso
 273     =>>>ucs@unicode
 274     ==ucs@iso
 275     ==ucs@unicode
 276     ==gb2312
 277     ==ks-x1001
 278     ==cns11643-1
 279     ==cns11643-2
 280     ==cns11643-3
 281     ==gt
 282     ==jis-x0208@1990
 283     ==gt-k
 284     =ucs@iso
 285     =ucs@unicode
 286     =ucs@cns
 287     =>>big5-cdp
 288     =>>gt-k
 289     =+>gt
 290     =>gt
 291     =>big5-cdp
 292     =>daikanwa
 293     =>daikanwa/ho
 294     =>cns11643-7
 295     =big5
 296     =big5-eten
 297     =>gt-k
 298     =zinbun-oracle
 299     =>zinbun-oracle
 300     =ruimoku-v6
 301     =>>ruimoku-v6
 302     =jef-china3
 303     =shinjigen
 304     =big5-cdp-var-3
 305     =big5-cdp-var-5))
 306
 307
 308 ;;; @ char-db formatters
 309 ;;;
 310
 311 (defun char-db-make-char-spec (char)
 312   (let (ret char-spec)
 313     (cond ((characterp char)
 314            (cond ((and (setq ret (encode-char char '=ucs 'defined-only))
 315                        (not (and (<= #xE000 ret)(<= ret #xF8FF))))
 316                   (setq char-spec (list (cons '=ucs ret)))
 317                   (cond ((setq ret (get-char-attribute char 'name))
 318                          (setq char-spec (cons (cons 'name ret) char-spec))
 319                          )
 320                         ((setq ret (get-char-attribute char 'name*))
 321                          (setq char-spec (cons (cons 'name* ret) char-spec))
 322                          ))
 323                   )
 324                  ((encode-char char '=mj 'defined-only)
 325                   (setq char-spec nil)
 326                   (dolist (ccs (charset-list))
 327                     (if (and (or (eq ccs '=mj)
 328                                  ;; (eq (charset-property ccs 'iso-ir) 177)
 329                                  (string-match "=ucs@" (symbol-name ccs))
 330                                  )
 331                              (setq ccs (charset-name ccs))
 332                              (null (assq ccs char-spec))
 333                              (setq ret (encode-char char ccs 'defined-only)))
 334                         (setq char-spec (cons (cons ccs ret) char-spec))))
 335                   )
 336                  ((encode-char char '==mj 'defined-only)
 337                   (setq char-spec nil)
 338                   (dolist (ccs (charset-list))
 339                     (if (and (or (eq ccs '==mj)
 340                                  ;; (eq (charset-property ccs 'iso-ir) 177)
 341                                  (string-match "=ucs@" (symbol-name ccs))
 342                                  )
 343                              (setq ccs (charset-name ccs))
 344                              (null (assq ccs char-spec))
 345                              (setq ret (encode-char char ccs 'defined-only)))
 346                         (setq char-spec (cons (cons ccs ret) char-spec))))
 347                   )
 348                  ((encode-char char '=adobe-japan1 'defined-only)
 349                   (setq char-spec nil)
 350                   (dolist (ccs (charset-list))
 351                     (if (and (or (memq ccs
 352                                        '(=adobe-japan1-0
 353                                          =adobe-japan1-1
 354                                          =adobe-japan1-2
 355                                          =adobe-japan1-3
 356                                          =adobe-japan1-4
 357                                          =adobe-japan1-5
 358                                          =adobe-japan1-6
 359                                          ))
 360                                  ;; (eq (charset-property ccs 'iso-ir) 177)
 361                                  (string-match "=ucs@" (symbol-name ccs))
 362                                  )
 363                              (setq ccs (charset-name ccs))
 364                              (null (assq ccs char-spec))
 365                              (setq ret (encode-char char ccs 'defined-only)))
 366                         (setq char-spec (cons (cons ccs ret) char-spec))))
 367                   )
 368                  ((encode-char char '==adobe-japan1 'defined-only)
 369                   (setq char-spec nil)
 370                   (dolist (ccs (charset-list))
 371                     (if (and (or (memq ccs
 372                                        '(==adobe-japan1-0
 373                                          ==adobe-japan1-1
 374                                          ==adobe-japan1-2
 375                                          ==adobe-japan1-3
 376                                          ==adobe-japan1-4
 377                                          ==adobe-japan1-5
 378                                          ==adobe-japan1-6
 379                                          ))
 380                                  ;; (eq (charset-property ccs 'iso-ir) 177)
 381                                  (string-match "=ucs@" (symbol-name ccs))
 382                                  )
 383                              (setq ccs (charset-name ccs))
 384                              (null (assq ccs char-spec))
 385                              (setq ret (encode-char char ccs 'defined-only)))
 386                         (setq char-spec (cons (cons ccs ret) char-spec))))
 387                   )
 388                  ((setq ret
 389                         (catch 'tag
 390                           (let ((rest char-db-coded-charset-priority-list)
 391                                 ccs)
 392                             (while rest
 393                               (setq ccs (charset-name
 394                                          (find-charset (car rest))))
 395                               (if (setq ret
 396                                         (encode-char char ccs
 397                                                      'defined-only))
 398                                   (throw 'tag (cons ccs ret)))
 399                               (setq rest (cdr rest))))))
 400                   (setq char-spec (list ret))
 401                   (dolist (ccs (delq (car ret) (charset-list)))
 402                     (if (and (or (charset-iso-final-char ccs)
 403                                  (memq ccs
 404                                        '(=daikanwa
 405                                          =daikanwa@rev2
 406                                          ;; =gt-k
 407                                          =jis-x0208@1997
 408                                          ))
 409                                  (eq (charset-property ccs 'iso-ir) 177)
 410                                  ;; (string-match "=ucs@" (symbol-name ccs))
 411                                  )
 412                              (setq ccs (charset-name ccs))
 413                              (null (assq ccs char-spec))
 414                              (setq ret (encode-char char ccs 'defined-only)))
 415                         (setq char-spec (cons (cons ccs ret) char-spec))))
 416                   (if (null char-spec)
 417                       (setq char-spec (split-char char)))
 418                   (cond ((setq ret (get-char-attribute char 'name))
 419                          (setq char-spec (cons (cons 'name ret) char-spec))
 420                          )
 421                         ((setq ret (get-char-attribute char 'name*))
 422                          (setq char-spec (cons (cons 'name* ret) char-spec))
 423                          ))
 424                   )
 425                  ((setq ret (get-char-attribute
 426                              char 'ideographic-combination))
 427                   (setq char-spec
 428                         (cons (cons 'ideographic-combination ret)
 429                               char-spec))
 430                   ))
 431            char-spec)
 432           ((consp char)
 433            char))))
 434
 435 (defun char-db-insert-char-spec (char &optional readable column
 436                                       required-features)
 437   (unless column
 438     (setq column (current-column)))
 439   (let (char-spec temp-char)
 440     (setq char-spec (char-db-make-char-spec char))
 441     (unless (or (characterp char) ; char
 442                 (condition-case nil
 443                     (setq char (find-char char-spec))
 444                   (error nil)))
 445       ;; define temporary character
 446       ;;   Current implementation is dirty.
 447       (setq temp-char (define-char (cons '(ideograph-daikanwa . 0)
 448                                          char-spec)))
 449       (remove-char-attribute temp-char 'ideograph-daikanwa)
 450       (setq char temp-char))
 451     (insert-char-attributes char
 452                             readable
 453                             (union (mapcar #'car char-spec)
 454                                    required-features)
 455                             nil 'for-sub-node)
 456     (when temp-char
 457       ;; undefine temporary character
 458       ;;   Current implementation is dirty.
 459       (setq char-spec (char-attribute-alist temp-char))
 460       (while char-spec
 461         (remove-char-attribute temp-char (car (car char-spec)))
 462         (setq char-spec (cdr char-spec))))))
 463
 464 (defun char-db-insert-alist (alist &optional readable column)
 465   (unless column
 466     (setq column (current-column)))
 467   (let ((line-breaking
 468          (concat "\n" (make-string (1+ column) ?\ )))
 469         name value
 470         ret al ; cal
 471         key
 472         lbs cell rest separator)
 473     (insert "(")
 474     (while alist
 475       (setq name (car (car alist))
 476             value (cdr (car alist)))
 477       (cond ((eq name 'char)
 478              (insert "(char . ")
 479              (if (setq ret (condition-case nil
 480                                (find-char value)
 481                              (error nil)))
 482                  (progn
 483                    (setq al nil
 484                          ;; cal nil
 485                          )
 486                    (while value
 487                      (setq key (car (car value)))
 488                      ;; (if (find-charset key)
 489                      ;;     (setq cal (cons key cal))
 490                      (setq al (cons key al))
 491                      ;; )
 492                      (setq value (cdr value)))
 493                    (insert-char-attributes ret
 494                                            readable
 495                                            (or al 'none) ; cal
 496                                            nil 'for-sub-node))
 497                (insert (prin1-to-string value)))
 498              (insert ")")
 499              (insert line-breaking))
 500             ((consp value)
 501              (insert (format "(%-18s " name))
 502              (setq lbs (concat "\n" (make-string (current-column) ?\ )))
 503              (while (consp value)
 504                (setq cell (car value))
 505                (if (and (consp cell)
 506                         (consp (car cell))
 507                         (setq ret (condition-case nil
 508                                       (find-char cell)
 509                                     (error nil)))
 510                         )
 511                    (progn
 512                      (setq rest cell
 513                            al nil
 514                            ;; cal nil
 515                            )
 516                      (while rest
 517                        (setq key (car (car rest)))
 518                        ;; (if (find-charset key)
 519                        ;;     (setq cal (cons key cal))
 520                        (setq al (cons key al))
 521                        ;; )
 522                        (setq rest (cdr rest)))
 523                      (if separator
 524                          (insert lbs))
 525                      (insert-char-attributes ret
 526                                              readable
 527                                              al ; cal
 528                                              nil 'for-sub-node)
 529                      (setq separator lbs))
 530                  (if separator
 531                      (insert separator))
 532                  (insert (prin1-to-string cell))
 533                  (setq separator " "))
 534                (setq value (cdr value)))
 535              (insert ")")
 536              (insert line-breaking))
 537             (t
 538              (insert (format "(%-18s . %S)%s"
 539                              name value
 540                              line-breaking))))
 541       (setq alist (cdr alist))))
 542   (insert ")"))
 543
 544 (defun char-db-insert-char-reference (plist &optional readable column)
 545   (unless column
 546     (setq column (current-column)))
 547   (let ((line-breaking
 548          (concat "\n" (make-string (1+ column) ?\ )))
 549         (separator "")
 550         name value)
 551     (insert "(")
 552     (while plist
 553       (setq name (pop plist))
 554       (setq value (pop plist))
 555       (cond ((eq name :char)
 556              (insert separator)
 557              (insert ":char\t")
 558              (cond ((numberp value)
 559                     (setq value (decode-char '=ucs value)))
 560                    ;; ((consp value)
 561                    ;;  (setq value (or (find-char value)
 562                    ;;                  value)))
 563                    )
 564              (char-db-insert-char-spec value readable)
 565              (insert line-breaking)
 566              (setq separator ""))
 567             ((eq name :radical)
 568              (insert (format "%s%s\t%d ; %c%s"
 569                              separator
 570                              name value
 571                              (ideographic-radical value)
 572                              line-breaking))
 573              (setq separator ""))
 574             (t
 575              (insert (format "%s%s\t%S" separator name value))
 576              (setq separator line-breaking)))
 577       ))
 578   (insert ")"))
 579
 580 (defun char-db-decode-isolated-char (ccs code-point)
 581   (let (ret)
 582     (setq ret
 583           (cond ((eq ccs 'arabic-iso8859-6)
 584                  (decode-char ccs code-point))
 585                 ;; ((eq ccs '=gt)
 586                 ;;  (decode-builtin-char '==gt code-point))
 587                 ((and (memq ccs '(=gt-pj-1
 588                                   =gt-pj-2
 589                                   =gt-pj-3
 590                                   =gt-pj-4
 591                                   =gt-pj-5
 592                                   =gt-pj-6
 593                                   =gt-pj-7
 594                                   =gt-pj-8
 595                                   =gt-pj-9
 596                                   =gt-pj-10
 597                                   =gt-pj-11))
 598                       (setq ret (decode-char ccs code-point))
 599                       (setq ret (encode-char ret '=gt 'defined-only)))
 600                  (decode-builtin-char '=gt ret))
 601                 (t
 602                  (decode-builtin-char ccs code-point))))
 603     (cond ((and (<= 0 (char-int ret))
 604                 (<= (char-int ret) #x1F))
 605            (decode-char '=ucs (+ #x2400 (char-int ret))))
 606           ((= (char-int ret) #x7F)
 607            ?\u2421)
 608           (t ret))))
 609
 610 (defvar char-db-convert-obsolete-format t)
 611
 612 (defun char-db-insert-ccs-feature (name value line-breaking)
 613   (cond
 614    ((integerp value)
 615     (insert
 616      (format
 617       (cond
 618        ((memq name '(==shinjigen
 619                      =shinjigen
 620                      =shinjigen@1ed ==shinjigen@1ed
 621                      =shinjigen@rev ==shinjigen@rev
 622                      =shinjigen/+p@rev ==shinjigen/+p@rev
 623                      ==daikanwa/ho
 624                      =daikanwa/ho =>>daikanwa/ho =>daikanwa/ho))
 625         "(%-18s .  %04d)\t; %c")
 626        ((eq name '=shinjigen@1ed/24pr)
 627         "(%-18s . %04d)\t; %c")
 628        ((or (memq name '(==daikanwa
 629                          =daikanwa =>>daikanwa =>daikanwa
 630                          =daikanwa@rev1 =daikanwa@rev2
 631                          =daikanwa/+p ==daikanwa/+p =>>daikanwa/+p
 632                          =daikanwa/+2p =>>daikanwa/+2p
 633                          =gt ==gt ; =>>>gt
 634                          =>>gt =+>gt =>gt
 635                          =gt-k ==gt-k =>>gt-k =>gt-k
 636                          =adobe-japan1-0 ==adobe-japan1-0 ; =>>>adobe-japan1-0
 637                          =adobe-japan1-1 ==adobe-japan1-1 ; =>>>adobe-japan1-1
 638                          =adobe-japan1-2 ==adobe-japan1-2 ; =>>>adobe-japan1-2
 639                          =adobe-japan1-3 ==adobe-japan1-3 ; =>>>adobe-japan1-3
 640                          =adobe-japan1-4 ==adobe-japan1-4 ; =>>>adobe-japan1-4
 641                          =adobe-japan1-5 ==adobe-japan1-5 ; =>>>adobe-japan1-5
 642                          =adobe-japan1-6 ==adobe-japan1-6 ; =>>>adobe-japan1-6
 643                          =>>adobe-japan1-0 =+>adobe-japan1-0
 644                          =>>adobe-japan1-1 =+>adobe-japan1-1
 645                          =>>adobe-japan1-2 =+>adobe-japan1-2
 646                          =>>adobe-japan1-3 =+>adobe-japan1-3
 647                          =>>adobe-japan1-4 =+>adobe-japan1-4
 648                          =>>adobe-japan1-5 =+>adobe-japan1-5
 649                          =>>adobe-japan1-6 =+>adobe-japan1-6
 650                          =cbeta =>>cbeta
 651                          =zinbun-oracle =>zinbun-oracle))
 652             ;; (string-match "^=adobe-" (symbol-name name))
 653             )
 654         "(%-18s . %05d)\t; %c")
 655        ((memq name '(=hanyo-denshi/ks
 656                      ==hanyo-denshi/ks ; =>>>hanyo-denshi/ks
 657                      =>>hanyo-denshi/ks
 658                      =koseki
 659                      =mj ==mj =>>mj
 660                      =zihai mojikyo))
 661         "(%-18s . %06d)\t; %c")
 662        ((>= (charset-dimension name) 2)
 663         "(%-18s . #x%04X)\t; %c")
 664        (t
 665         "(%-18s . #x%02X)\t; %c"))
 666       name
 667       (if (= (charset-iso-graphic-plane name) 1)
 668           (logior value
 669                   (cond ((= (charset-dimension name) 1)
 670                          #x80)
 671                         ((= (charset-dimension name) 2)
 672                          #x8080)
 673                         ((= (charset-dimension name) 3)
 674                          #x808080)
 675                         (t 0)))
 676         value)
 677       (char-db-decode-isolated-char name value)))
 678     (if (and (= (charset-chars name) 94)
 679              (= (charset-dimension name) 2))
 680         (insert (format " [%02d-%02d]"
 681                         (- (lsh value -8) 32)
 682                         (- (logand value 255) 32))))
 683     )
 684    (t
 685     (insert (format "(%-18s . %s)" name value))
 686     ))
 687   (insert line-breaking))
 688
 689 (defun char-db-insert-relation-feature (char name value line-breaking
 690                                              ccss readable)
 691   (insert (format "(%-18s%s " name line-breaking))
 692   (let ((lbs (concat "\n" (make-string (current-column) ?\ )))
 693         separator cell sources required-features
 694         ret)
 695     (while (consp value)
 696       (setq cell (car value))
 697       (if (integerp cell)
 698           (setq cell (decode-char '=ucs cell)))
 699       (cond
 700        ((eq name '->subsumptive)
 701         (when (or (not (some (lambda (atr)
 702                                (get-char-attribute cell atr))
 703                              char-db-ignored-attributes))
 704                   (some (lambda (ccs)
 705                           (encode-char cell ccs 'defined-only))
 706                         ccss))
 707           (if separator
 708               (insert lbs))
 709           (let ((char-db-ignored-attributes
 710                  (cons '<-subsumptive
 711                        char-db-ignored-attributes)))
 712             (insert-char-attributes cell readable nil nil 'for-sub-node))
 713           (setq separator lbs))
 714         )
 715        ((characterp cell)
 716         (setq sources
 717               (get-char-attribute
 718                char (intern (format "%s*sources" name))))
 719         (setq required-features nil)
 720         (dolist (source sources)
 721           (cond
 722            ((memq source '(JP
 723                            JP/Jouyou
 724                            shinjigen shinjigen@1ed shinjigen@rev))
 725             (setq required-features
 726                   (union required-features
 727                          '(=jis-x0208
 728                            =jis-x0208@1990
 729                            =jis-x0213-1@2000
 730                            =jis-x0213-1@2004
 731                            =jis-x0213-2
 732                            =jis-x0212
 733                            =jis-x0208@1983
 734                            =jis-x0208@1978
 735                            =shinjigen
 736                            =shinjigen@1ed
 737                            =shinjigen@rev
 738                            =shinjigen/+p@rev))))
 739            ((eq source 'CN)
 740             (setq required-features
 741                   (union required-features
 742                          '(=gb2312
 743                            =gb12345
 744                            =iso-ir165)))))
 745           (cond
 746            ((find-charset (setq ret (intern (format "=%s" source))))
 747             (setq required-features
 748                   (cons ret required-features)))
 749            (t (setq required-features
 750                     (cons source required-features)))))
 751         (cond ((string-match "@JP" (symbol-name name))
 752                (setq required-features
 753                      (union required-features
 754                             '(=jis-x0208
 755                               =jis-x0208@1990
 756                               =jis-x0213-1-2000
 757                               =jis-x0213-2-2000
 758                               =jis-x0212
 759                               =jis-x0208@1983
 760                               =jis-x0208@1978))))
 761               ((string-match "@CN" (symbol-name name))
 762                (setq required-features
 763                      (union required-features
 764                             '(=gb2312
 765                               =gb12345
 766                               =iso-ir165)))))
 767         (if separator
 768             (insert lbs))
 769         (if readable
 770             (insert (format "%S" cell))
 771           (char-db-insert-char-spec cell readable
 772                                     nil
 773                                     required-features))
 774         (setq separator lbs))
 775        ((consp cell)
 776         (if separator
 777             (insert lbs))
 778         (if (consp (car cell))
 779             (char-db-insert-char-spec cell readable)
 780           (char-db-insert-char-reference cell readable))
 781         (setq separator lbs))
 782        (t
 783         (if separator
 784             (insert separator))
 785         (insert (prin1-to-string cell))
 786         (setq separator " ")))
 787       (setq value (cdr value)))
 788     (insert ")")
 789     (insert line-breaking)))
 790
 791 (defun insert-char-attributes (char &optional readable attributes column
 792                                     for-sub-node)
 793   (unless column
 794     (setq column (current-column)))
 795   (let (name value ; has-long-ccs-name
 796         rest
 797         radical strokes
 798         (line-breaking
 799          (concat "\n" (make-string (1+ column) ?\ )))
 800         lbs cell separator ret
 801         key al cal
 802         dest-ccss ; sources required-features
 803         ccss)
 804     (let (atr-d)
 805       (setq attributes
 806             (sort (if attributes
 807                       (if (consp attributes)
 808                           (progn
 809                             (dolist (name attributes)
 810                               (unless (memq name char-db-ignored-attributes)
 811                                 (if (find-charset name)
 812                                     (push name ccss))
 813                                 (push name atr-d)))
 814                             atr-d))
 815                     (dolist (name (char-attribute-list))
 816                       (unless (memq name char-db-ignored-attributes)
 817                         (if (find-charset name)
 818                             (push name ccss))
 819                         (push name atr-d)))
 820                     atr-d)
 821                   #'char-attribute-name<)))
 822     (insert "(")
 823     (when (memq '<-subsumptive attributes)
 824       (when (or readable (not for-sub-node))
 825         (when (setq value (get-char-attribute char '<-subsumptive))
 826           (char-db-insert-relation-feature char '<-subsumptive value
 827                                            line-breaking
 828                                            ccss readable)))
 829       (setq attributes (delq '<-subsumptive attributes)))
 830     (when (and (memq '<-denotational attributes)
 831                (setq value (get-char-attribute char '<-denotational)))
 832       (char-db-insert-relation-feature char '<-denotational value
 833                                        line-breaking
 834                                        ccss readable)
 835       (setq attributes (delq '<-denotational attributes)))
 836     (when (and (memq 'name attributes)
 837                (setq value (get-char-attribute char 'name)))
 838       (insert (format
 839                (if (> (+ (current-column) (length value)) 48)
 840                    "(name . %S)%s"
 841                  "(name               . %S)%s")
 842                value line-breaking))
 843       (setq attributes (delq 'name attributes))
 844       )
 845     (when (and (memq 'name* attributes)
 846                (setq value (get-char-attribute char 'name*)))
 847       (insert (format
 848                (if (> (+ (current-column) (length value)) 48)
 849                    "(name* . %S)%s"
 850                  "(name*              . %S)%s")
 851                value line-breaking))
 852       (setq attributes (delq 'name* attributes))
 853       )
 854     (when (and (memq 'script attributes)
 855                (setq value (get-char-attribute char 'script)))
 856       (insert (format "(script\t\t%s)%s"
 857                       (mapconcat (function prin1-to-string)
 858                                  value " ")
 859                       line-breaking))
 860       (setq attributes (delq 'script attributes))
 861       )
 862     (dolist (name '(=>ucs =>ucs*))
 863       (when (and (memq name attributes)
 864                  (setq value (get-char-attribute char name)))
 865         (insert (format "(%-18s . #x%04X)\t; %c%s"
 866                         name value (decode-char '=ucs value)
 867                         line-breaking))
 868         (setq attributes (delq name attributes))))
 869     (dolist (name '(=>ucs@gb =>ucs@big5))
 870       (when (and (memq name attributes)
 871                  (setq value (get-char-attribute char name)))
 872         (insert (format "(%-18s . #x%04X)\t; %c%s"
 873                         name value
 874                         (decode-char (intern
 875                                       (concat "="
 876                                               (substring
 877                                                (symbol-name name) 2)))
 878                                      value)
 879                         line-breaking))
 880         (setq attributes (delq name attributes))
 881         ))
 882     ;; (dolist (name '(=>daikanwa))
 883     ;;   (when (and (memq name attributes)
 884     ;;              (setq value (get-char-attribute char name)))
 885     ;;     (insert
 886     ;;      (if (integerp value)
 887     ;;          (format "(%-18s . %05d)\t; %c%s"
 888     ;;                  name value (decode-char '=daikanwa value)
 889     ;;                  line-breaking)
 890     ;;        (format "(%-18s %s)\t; %c%s"
 891     ;;                name
 892     ;;                (mapconcat (function prin1-to-string)
 893     ;;                           value " ")
 894     ;;                (char-representative-of-daikanwa char)
 895     ;;                line-breaking)))
 896     ;;     (setq attributes (delq name attributes))))
 897     (when (and (memq 'general-category attributes)
 898                (setq value (get-char-attribute char 'general-category)))
 899       (insert (format
 900                "(general-category\t%s) ; %s%s"
 901                (mapconcat (lambda (cell)
 902                             (format "%S" cell))
 903                           value " ")
 904                (cond ((rassoc value unidata-normative-category-alist)
 905                       "Normative Category")
 906                      ((rassoc value unidata-informative-category-alist)
 907                       "Informative Category")
 908                      (t
 909                       "Unknown Category"))
 910                line-breaking))
 911       (setq attributes (delq 'general-category attributes))
 912       )
 913     (when (and (memq 'bidi-category attributes)
 914                (setq value (get-char-attribute char 'bidi-category)))
 915       (insert (format "(bidi-category\t. %S)%s"
 916                       value
 917                       line-breaking))
 918       (setq attributes (delq 'bidi-category attributes))
 919       )
 920     (unless (or (not (memq 'mirrored attributes))
 921                 (eq (setq value (get-char-attribute char 'mirrored 'empty))
 922                     'empty))
 923       (insert (format "(mirrored\t\t. %S)%s"
 924                       value
 925                       line-breaking))
 926       (setq attributes (delq 'mirrored attributes))
 927       )
 928     (cond
 929      ((and (memq 'decimal-digit-value attributes)
 930            (setq value (get-char-attribute char 'decimal-digit-value)))
 931       (insert (format "(decimal-digit-value . %S)%s"
 932                       value
 933                       line-breaking))
 934       (setq attributes (delq 'decimal-digit-value attributes))
 935       (when (and (memq 'digit-value attributes)
 936                  (setq value (get-char-attribute char 'digit-value)))
 937         (insert (format "(digit-value\t . %S)%s"
 938                         value
 939                         line-breaking))
 940         (setq attributes (delq 'digit-value attributes))
 941         )
 942       (when (and (memq 'numeric-value attributes)
 943                  (setq value (get-char-attribute char 'numeric-value)))
 944         (insert (format "(numeric-value\t . %S)%s"
 945                         value
 946                         line-breaking))
 947         (setq attributes (delq 'numeric-value attributes))
 948         )
 949       )
 950      (t
 951       (when (and (memq 'digit-value attributes)
 952                  (setq value (get-char-attribute char 'digit-value)))
 953         (insert (format "(digit-value\t. %S)%s"
 954                         value
 955                         line-breaking))
 956         (setq attributes (delq 'digit-value attributes))
 957         )
 958       (when (and (memq 'numeric-value attributes)
 959                  (setq value (get-char-attribute char 'numeric-value)))
 960         (insert (format "(numeric-value\t. %S)%s"
 961                         value
 962                         line-breaking))
 963         (setq attributes (delq 'numeric-value attributes))
 964         )))
 965     (when (and (memq 'iso-10646-comment attributes)
 966                (setq value (get-char-attribute char 'iso-10646-comment)))
 967       (insert (format "(iso-10646-comment\t. %S)%s"
 968                       value
 969                       line-breaking))
 970       (setq attributes (delq 'iso-10646-comment attributes))
 971       )
 972     (when (and (memq 'morohashi-daikanwa attributes)
 973                (setq value (get-char-attribute char 'morohashi-daikanwa)))
 974       (insert (format "(morohashi-daikanwa\t%s)%s"
 975                       (mapconcat (function prin1-to-string) value " ")
 976                       line-breaking))
 977       (setq attributes (delq 'morohashi-daikanwa attributes))
 978       )
 979     (setq radical nil
 980           strokes nil)
 981     (when (and (memq 'ideographic-radical attributes)
 982                (setq value (get-char-attribute char 'ideographic-radical)))
 983       (setq radical value)
 984       (insert (format "(ideographic-radical . %S)\t; %c%s"
 985                       radical
 986                       (ideographic-radical radical)
 987                       line-breaking))
 988       (setq attributes (delq 'ideographic-radical attributes))
 989       )
 990     (when (and (memq 'shuowen-radical attributes)
 991                (setq value (get-char-attribute char 'shuowen-radical)))
 992       (insert (format "(shuowen-radical\t. %S)\t; %c%s"
 993                       value
 994                       (shuowen-radical value)
 995                       line-breaking))
 996       (setq attributes (delq 'shuowen-radical attributes))
 997       )
 998     (let (key)
 999       (dolist (domain
1000                (append
1001                 char-db-feature-domains
1002                 (let (dest domain)
1003                   (dolist (feature (char-attribute-list))
1004                     (setq feature (symbol-name feature))
1005                     (when (string-match
1006                            "\\(radical\\|strokes\\)@\\([^@*]+\\)\\(\\*\\|$\\)"
1007                            feature)
1008                       (setq domain (intern (match-string 2 feature)))
1009                      (unless (memq domain dest)
1010                        (setq dest (cons domain dest)))))
1011                   (sort dest #'string<))))
1012         (setq key (intern (format "%s@%s" 'ideographic-radical domain)))
1013         (when (and (memq key attributes)
1014                    (setq value (get-char-attribute char key)))
1015           (setq radical value)
1016           (insert (format "(%s . %S)\t; %c%s"
1017                           key
1018                           radical
1019                           (ideographic-radical radical)
1020                           line-breaking))
1021           (setq attributes (delq key attributes))
1022           )
1023         (setq key (intern (format "%s@%s" 'ideographic-strokes domain)))
1024         (when (and (memq key attributes)
1025                    (setq value (get-char-attribute char key)))
1026           (setq strokes value)
1027           (insert (format "(%s . %S)%s"
1028                           key
1029                           strokes
1030                           line-breaking))
1031           (setq attributes (delq key attributes))
1032           )
1033         (setq key (intern (format "%s@%s" 'total-strokes domain)))
1034         (when (and (memq key attributes)
1035                    (setq value (get-char-attribute char key)))
1036           (insert (format "(%s       . %S)%s"
1037                           key
1038                           value
1039                           line-breaking))
1040           (setq attributes (delq key attributes))
1041           )
1042         (dolist (feature '(ideographic-radical
1043                            ideographic-strokes
1044                            total-strokes))
1045           (setq key (intern (format "%s@%s*sources" feature domain)))
1046           (when (and (memq key attributes)
1047                      (setq value (get-char-attribute char key)))
1048             (insert (format "(%s%s" key line-breaking))
1049             (dolist (cell value)
1050               (insert (format " %s" cell)))
1051             (insert ")")
1052             (insert line-breaking)
1053             (setq attributes (delq key attributes))
1054             ))
1055         ))
1056     (when (and (memq 'ideographic-strokes attributes)
1057                (setq value (get-char-attribute char 'ideographic-strokes)))
1058       (setq strokes value)
1059       (insert (format "(ideographic-strokes . %S)%s"
1060                       strokes
1061                       line-breaking))
1062       (setq attributes (delq 'ideographic-strokes attributes))
1063       )
1064     (when (and (memq 'kangxi-radical attributes)
1065                (setq value (get-char-attribute char 'kangxi-radical)))
1066       (unless (eq value radical)
1067         (insert (format "(kangxi-radical\t . %S)\t; %c%s"
1068                         value
1069                         (ideographic-radical value)
1070                         line-breaking))
1071         (or radical
1072             (setq radical value)))
1073       (setq attributes (delq 'kangxi-radical attributes))
1074       )
1075     (when (and (memq 'kangxi-strokes attributes)
1076                (setq value (get-char-attribute char 'kangxi-strokes)))
1077       (unless (eq value strokes)
1078         (insert (format "(kangxi-strokes\t . %S)%s"
1079                         value
1080                         line-breaking))
1081         (or strokes
1082             (setq strokes value)))
1083       (setq attributes (delq 'kangxi-strokes attributes))
1084       )
1085     (when (and (memq 'japanese-radical attributes)
1086                (setq value (get-char-attribute char 'japanese-radical)))
1087       (unless (eq value radical)
1088         (insert (format "(japanese-radical\t . %S)\t; %c%s"
1089                         value
1090                         (ideographic-radical value)
1091                         line-breaking))
1092         (or radical
1093             (setq radical value)))
1094       (setq attributes (delq 'japanese-radical attributes))
1095       )
1096     (when (and (memq 'japanese-strokes attributes)
1097                (setq value (get-char-attribute char 'japanese-strokes)))
1098       (unless (eq value strokes)
1099         (insert (format "(japanese-strokes\t . %S)%s"
1100                         value
1101                         line-breaking))
1102         (or strokes
1103             (setq strokes value)))
1104       (setq attributes (delq 'japanese-strokes attributes))
1105       )
1106     (when (and (memq 'cns-radical attributes)
1107                (setq value (get-char-attribute char 'cns-radical)))
1108       (insert (format "(cns-radical\t . %S)\t; %c%s"
1109                       value
1110                       (ideographic-radical value)
1111                       line-breaking))
1112       (setq attributes (delq 'cns-radical attributes))
1113       )
1114     (when (and (memq 'cns-strokes attributes)
1115                (setq value (get-char-attribute char 'cns-strokes)))
1116       (unless (eq value strokes)
1117         (insert (format "(cns-strokes\t . %S)%s"
1118                         value
1119                         line-breaking))
1120         (or strokes
1121             (setq strokes value)))
1122       (setq attributes (delq 'cns-strokes attributes))
1123       )
1124     ;; (when (and (memq 'shinjigen-1-radical attributes)
1125     ;;            (setq value (get-char-attribute char 'shinjigen-1-radical)))
1126     ;;   (unless (eq value radical)
1127     ;;     (insert (format "(shinjigen-1-radical . %S)\t; %c%s"
1128     ;;                     value
1129     ;;                     (ideographic-radical value)
1130     ;;                     line-breaking))
1131     ;;     (or radical
1132     ;;         (setq radical value)))
1133     ;;   (setq attributes (delq 'shinjigen-1-radical attributes))
1134     ;;   )
1135     (when (and (memq 'ideographic- attributes)
1136                (setq value (get-char-attribute char 'ideographic-)))
1137       (insert "(ideographic-       ")
1138       (setq lbs (concat "\n" (make-string (current-column) ?\ ))
1139             separator nil)
1140       (while (consp value)
1141         (setq cell (car value))
1142         (if (integerp cell)
1143             (setq cell (decode-char '=ucs cell)))
1144         (cond ((characterp cell)
1145                (if separator
1146                    (insert lbs))
1147                (if readable
1148                    (insert (format "%S" cell))
1149                  (char-db-insert-char-spec cell readable))
1150                (setq separator lbs))
1151               ((consp cell)
1152                (if separator
1153                    (insert lbs))
1154                (if (consp (car cell))
1155                    (char-db-insert-char-spec cell readable)
1156                  (char-db-insert-char-reference cell readable))
1157                (setq separator lbs))
1158               (t
1159                (if separator
1160                    (insert separator))
1161                (insert (prin1-to-string cell))
1162                (setq separator " ")))
1163         (setq value (cdr value)))
1164       (insert ")")
1165       (insert line-breaking)
1166       (setq attributes (delq 'ideographic- attributes)))
1167     (when (and (memq 'total-strokes attributes)
1168                (setq value (get-char-attribute char 'total-strokes)))
1169       (insert (format "(total-strokes       . %S)%s"
1170                       value
1171                       line-breaking))
1172       (setq attributes (delq 'total-strokes attributes))
1173       )
1174     (when (and (memq '->ideograph attributes)
1175                (setq value (get-char-attribute char '->ideograph)))
1176       (insert (format "(->ideograph\t%s)%s"
1177                       (mapconcat (lambda (code)
1178                                    (cond ((symbolp code)
1179                                           (symbol-name code))
1180                                          ((integerp code)
1181                                           (format "#x%04X" code))
1182                                          (t
1183                                           (format "%s %S"
1184                                                   line-breaking code))))
1185                                  value " ")
1186                       line-breaking))
1187       (setq attributes (delq '->ideograph attributes))
1188       )
1189     ;; (when (and (memq '->decomposition attributes)
1190     ;;            (setq value (get-char-attribute char '->decomposition)))
1191     ;;   (insert (format "(->decomposition\t%s)%s"
1192     ;;                   (mapconcat (lambda (code)
1193     ;;                                (cond ((symbolp code)
1194     ;;                                       (symbol-name code))
1195     ;;                                      ((characterp code)
1196     ;;                                       (if readable
1197     ;;                                           (format "%S" code)
1198     ;;                                         (format "#x%04X"
1199     ;;                                                 (char-int code))
1200     ;;                                         ))
1201     ;;                                      ((integerp code)
1202     ;;                                       (format "#x%04X" code))
1203     ;;                                      (t
1204     ;;                                       (format "%s%S" line-breaking code))))
1205     ;;                              value " ")
1206     ;;                   line-breaking))
1207     ;;   (setq attributes (delq '->decomposition attributes))
1208     ;;   )
1209     (if (equal (get-char-attribute char '->titlecase)
1210                (get-char-attribute char '->uppercase))
1211         (setq attributes (delq '->titlecase attributes)))
1212     (when (and (memq '->mojikyo attributes)
1213                (setq value (get-char-attribute char '->mojikyo)))
1214       (insert (format "(->mojikyo\t\t. %06d)\t; %c%s"
1215                       value (decode-char 'mojikyo value)
1216                       line-breaking))
1217       (setq attributes (delq '->mojikyo attributes))
1218       )
1219     (when (and (memq 'hanyu-dazidian-vol attributes)
1220                (setq value (get-char-attribute char 'hanyu-dazidian-vol)))
1221       (insert (format "(hanyu-dazidian-vol  . %d)%s"
1222                       value line-breaking))
1223       (setq attributes (delq 'hanyu-dazidian-vol attributes))
1224       )
1225     (when (and (memq 'hanyu-dazidian-page attributes)
1226                (setq value (get-char-attribute char 'hanyu-dazidian-page)))
1227       (insert (format "(hanyu-dazidian-page . %d)%s"
1228                       value line-breaking))
1229       (setq attributes (delq 'hanyu-dazidian-page attributes))
1230       )
1231     (when (and (memq 'hanyu-dazidian-char attributes)
1232                (setq value (get-char-attribute char 'hanyu-dazidian-char)))
1233       (insert (format "(hanyu-dazidian-char . %d)%s"
1234                       value line-breaking))
1235       (setq attributes (delq 'hanyu-dazidian-char attributes))
1236       )
1237     (unless readable
1238       (dolist (ignored '(composition
1239                          ->denotational <-subsumptive ->ucs-unified
1240                          ->ideographic-component-forms))
1241         (setq attributes (delq ignored attributes))))
1242     (while attributes
1243       (setq name (car attributes))
1244       (unless (eq (setq value (get-char-attribute char name 'value-is-empty))
1245                   'value-is-empty)
1246         (cond ((setq ret (find-charset name))
1247                (setq name (charset-name ret))
1248                (when (not (memq name dest-ccss))
1249                  (setq dest-ccss (cons name dest-ccss))
1250                  (char-db-insert-ccs-feature name value line-breaking))
1251                )
1252               ((string-match "^=>ucs@" (symbol-name name))
1253                (insert (format "(%-18s . #x%04X)\t; %c%s"
1254                                name value (decode-char '=ucs value)
1255                                line-breaking))
1256                )
1257               ((eq name 'jisx0208-1978/4X)
1258                (insert (format "(%-18s . #x%04X)%s"
1259                                name value
1260                                line-breaking))
1261                )
1262               ((and
1263                 (not readable)
1264                 (not (eq name '->subsumptive))
1265                 (not (eq name '->uppercase))
1266                 (not (eq name '->lowercase))
1267                 (not (eq name '->titlecase))
1268                 (not (eq name '->canonical))
1269                 (not (eq name '->Bopomofo))
1270                 (not (eq name '->mistakable))
1271                 (not (eq name '->ideographic-variants))
1272                 (null (get-char-attribute
1273                        char (intern (format "%s*sources" name))))
1274                 (not (string-match "\\*sources$" (symbol-name name)))
1275                 (null (get-char-attribute
1276                        char (intern (format "%s*note" name))))
1277                 (not (string-match "\\*note$" (symbol-name name)))
1278                 (or (eq name '<-identical)
1279                     (eq name '<-uppercase)
1280                     (eq name '<-lowercase)
1281                     (eq name '<-titlecase)
1282                     (eq name '<-canonical)
1283                     (eq name '<-ideographic-variants)
1284                     ;; (eq name '<-synonyms)
1285                     (string-match "^<-synonyms" (symbol-name name))
1286                     (eq name '<-mistakable)
1287                     (when (string-match "^->" (symbol-name name))
1288                       (cond
1289                        ((string-match "^->fullwidth" (symbol-name name))
1290                         (not (and (consp value)
1291                                   (characterp (car value))
1292                                   (encode-char
1293                                    (car value) '=ucs 'defined-only)))
1294                         )
1295                        (t)))
1296                     ))
1297                )
1298               ((or (eq name 'ideographic-structure)
1299                    (eq name 'ideographic-combination)
1300                    (eq name 'ideographic-)
1301                    (eq name '=decomposition)
1302                    (char-feature-base-name= '=decomposition name)
1303                    (char-feature-base-name= '=>decomposition name)
1304                    ;; (string-match "^=>*decomposition\\(@[^*]+\\)?$"
1305                    ;;               (symbol-name name))
1306                    (string-match "^\\(->\\|<-\\)[^*]*$" (symbol-name name))
1307                    (string-match "^\\(->\\|<-\\)[^*]*\\*sources$"
1308                                  (symbol-name name))
1309                    )
1310                (char-db-insert-relation-feature char name value
1311                                                 line-breaking
1312                                                 ccss readable))
1313               ((memq name '(ideograph=
1314                             original-ideograph-of
1315                             ancient-ideograph-of
1316                             vulgar-ideograph-of
1317                             wrong-ideograph-of
1318                             ;; simplified-ideograph-of
1319                             ideographic-variants
1320                             ;; ideographic-different-form-of
1321                             ))
1322                (insert (format "(%-18s%s " name line-breaking))
1323                (setq lbs (concat "\n" (make-string (current-column) ?\ ))
1324                      separator nil)
1325                (while (consp value)
1326                  (setq cell (car value))
1327                  (if (and (consp cell)
1328                           (consp (car cell)))
1329                      (progn
1330                        (if separator
1331                            (insert lbs))
1332                        (char-db-insert-alist cell readable)
1333                        (setq separator lbs))
1334                    (if separator
1335                        (insert separator))
1336                    (insert (prin1-to-string cell))
1337                    (setq separator " "))
1338                  (setq value (cdr value)))
1339                (insert ")")
1340                (insert line-breaking))
1341               ((consp value)
1342                (insert (format "(%-18s " name))
1343                (setq lbs (concat "\n" (make-string (current-column) ?\ ))
1344                      separator nil)
1345                (while (consp value)
1346                  (setq cell (car value))
1347                  (if (and (consp cell)
1348                           (consp (car cell))
1349                           (setq ret (condition-case nil
1350                                         (find-char cell)
1351                                       (error nil))))
1352                      (progn
1353                        (setq rest cell
1354                              al nil
1355                              cal nil)
1356                        (while rest
1357                          (setq key (car (car rest)))
1358                          (if (find-charset key)
1359                              (setq cal (cons key cal))
1360                            (setq al (cons key al)))
1361                          (setq rest (cdr rest)))
1362                        (if separator
1363                            (insert lbs))
1364                        (insert-char-attributes ret
1365                                                readable
1366                                                al ; cal
1367                                                nil 'for-sub-node)
1368                        (setq separator lbs))
1369                    (setq ret (prin1-to-string cell))
1370                    (if separator
1371                        (if (< (+ (current-column)
1372                                  (length ret)
1373                                  (length separator))
1374                               76)
1375                            (insert separator)
1376                          (insert lbs)))
1377                    (insert ret)
1378                    (setq separator " "))
1379                  (setq value (cdr value)))
1380                (insert ")")
1381                (insert line-breaking))
1382               (t
1383                (insert (format "(%-18s" name))
1384                (setq ret (prin1-to-string value))
1385                (unless (< (+ (current-column)
1386                              (length ret)
1387                              3)
1388                           76)
1389                  (insert line-breaking))
1390                (insert " . " ret ")" line-breaking)
1391                ;; (insert (format "(%-18s . %S)%s"
1392                ;;                 name value
1393                ;;                 line-breaking))
1394                )
1395               ))
1396       (setq attributes (cdr attributes)))
1397     (insert ")")))
1398
1399 (defun insert-char-data (char &optional readable
1400                               attributes)
1401   (save-restriction
1402     (narrow-to-region (point)(point))
1403     (insert "(define-char
1404   '")
1405     (insert-char-attributes char readable attributes)
1406     (insert ")\n")
1407     (goto-char (point-min))
1408     (while (re-search-forward "[ \t]+$" nil t)
1409       (replace-match ""))
1410     ;; from tabify.
1411     (goto-char (point-min))
1412     (while (re-search-forward "[ \t][ \t][ \t]*" nil t)
1413       (let ((column (current-column))
1414             (indent-tabs-mode t))
1415         (delete-region (match-beginning 0) (point))
1416         (indent-to column)))
1417     (goto-char (point-max))
1418     ;; (tabify (point-min)(point-max))
1419     ))
1420
1421 (defun insert-char-data-with-variant (char &optional printable
1422                                            no-ucs-unified
1423                                            script excluded-script)
1424   (insert-char-data char printable)
1425   (let ((variants (char-variants char))
1426         rest
1427         variant vs ret)
1428     (setq variants (sort variants #'<))
1429     (setq rest variants)
1430     (setq variants (cons char variants))
1431     (while rest
1432       (setq variant (car rest))
1433       (unless (get-char-attribute variant '<-subsumptive)
1434         (if (and (or (null script)
1435                      (null (setq vs (get-char-attribute variant 'script)))
1436                      (memq script vs))
1437                  (or (null excluded-script)
1438                      (null (setq vs (get-char-attribute variant 'script)))
1439                      (not (memq excluded-script vs))))
1440             (unless (and no-ucs-unified (get-char-attribute variant '=ucs))
1441               (insert-char-data variant printable)
1442               (if (setq ret (char-variants variant))
1443                   (while ret
1444                     (or (memq (car ret) variants)
1445                         ;; (get-char-attribute (car ret) '<-subsumptive)
1446                         (setq rest (nconc rest (list (car ret)))))
1447                     (setq ret (cdr ret)))))))
1448       (setq rest (cdr rest)))))
1449
1450 (defun insert-char-range-data (min max &optional script excluded-script)
1451   (let ((code min)
1452         char)
1453     (while (<= code max)
1454       (setq char (decode-char '=ucs code))
1455       (if (encode-char char '=ucs 'defined-only)
1456           (insert-char-data-with-variant char nil 'no-ucs-unified
1457                                          script excluded-script))
1458       (setq code (1+ code)))))
1459
1460 (defun write-char-range-data-to-file (min max file
1461                                           &optional script excluded-script)
1462   (let ((coding-system-for-write char-db-file-coding-system))
1463     (with-temp-buffer
1464       (insert (format ";; -*- coding: %s -*-\n"
1465                       char-db-file-coding-system))
1466       (insert-char-range-data min max script excluded-script)
1467       (write-region (point-min)(point-max) file))))
1468
1469 (defvar what-character-original-window-configuration)
1470
1471 ;;;###autoload
1472 (defun what-char-definition (char)
1473   (interactive (list (char-after)))
1474   (let ((buf (get-buffer-create "*Character Description*"))
1475         (the-buf (current-buffer))
1476         (win-conf (current-window-configuration)))
1477     (pop-to-buffer buf)
1478     (make-local-variable 'what-character-original-window-configuration)
1479     (setq what-character-original-window-configuration win-conf)
1480     (setq buffer-read-only nil)
1481     (erase-buffer)
1482     (condition-case err
1483         (progn
1484           (insert-char-data-with-variant char 'printable)
1485           (unless (char-attribute-alist char)
1486             (insert (format ";; = %c\n"
1487                             (let* ((rest (split-char char))
1488                                    (ccs (pop rest))
1489                                    (code (pop rest)))
1490                               (while rest
1491                                 (setq code (logior (lsh code 8)
1492                                                    (pop rest))))
1493                               (decode-char ccs code)))))
1494           ;; (char-db-update-comment)
1495           (set-buffer-modified-p nil)
1496           (view-mode the-buf (lambda (buf)
1497                                (set-window-configuration
1498                                 what-character-original-window-configuration)
1499                                ))
1500           (goto-char (point-min)))
1501       (error (progn
1502                (set-window-configuration
1503                 what-character-original-window-configuration)
1504                (signal (car err) (cdr err)))))))
1505
1506
1507 ;;; @ end
1508 ;;;
1509
1510 (provide 'char-db-util)
1511
1512 ;;; char-db-util.el ends here