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

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