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

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