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

   1 ;;; char-db-util.el --- Character Database utility -*- coding: utf-8-er; -*-
   2
   3 ;; Copyright (C) 1998,1999,2000,2001,2002,2003,2004,2005,2006,2007,
   4 ;;   2008,2009,2010,2011,2012,2013,2014,2015,2016,2017,2018 MORIOKA Tomohiko.
   5
   6 ;; Author: MORIOKA Tomohiko <tomo@kanji.zinbun.kyoto-u.ac.jp>
   7 ;; Keywords: CHISE, Character Database, ISO/IEC 10646, UCS, Unicode, MULE.
   8
   9 ;; This file is part of XEmacs CHISE.
  10
  11 ;; XEmacs CHISE is free software; you can redistribute it and/or
  12 ;; modify it under the terms of the GNU General Public License as
  13 ;; published by the Free Software Foundation; either version 2, or (at
  14 ;; your option) any later version.
  15
  16 ;; XEmacs CHISE is distributed in the hope that it will be useful,
  17 ;; but WITHOUT ANY WARRANTY; without even the implied warranty of
  18 ;; MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the GNU
  19 ;; General Public License for more details.
  20
  21 ;; You should have received a copy of the GNU General Public License
  22 ;; along with XEmacs CHISE; see the file COPYING.  If not, write to
  23 ;; the Free Software Foundation, Inc., 59 Temple Place - Suite 330,
  24 ;; Boston, MA 02111-1307, USA.
  25
  26 ;;; Code:
  27
  28 (require 'chise-subr)
  29 (require 'ideograph-subr)
  30
  31 (defconst unidata-normative-category-alist
  32   '(("Lu" letter        uppercase)
  33     ("Ll" letter        lowercase)
  34     ("Lt" letter        titlecase)
  35     ("Mn" mark          non-spacing)
  36     ("Mc" mark          spacing-combining)
  37     ("Me" mark          enclosing)
  38     ("Nd" number        decimal-digit)
  39     ("Nl" number        letter)
  40     ("No" number        other)
  41     ("Zs" separator     space)
  42     ("Zl" separator     line)
  43     ("Zp" separator     paragraph)
  44     ("Cc" other         control)
  45     ("Cf" other         format)
  46     ("Cs" other         surrogate)
  47     ("Co" other         private-use)
  48     ("Cn" other         not-assigned)))
  49
  50 (defconst unidata-informative-category-alist
  51   '(("Lm" letter        modifier)
  52     ("Lo" letter        other)
  53     ("Pc" punctuation   connector)
  54     ("Pd" punctuation   dash)
  55     ("Ps" punctuation   open)
  56     ("Pe" punctuation   close)
  57     ("Pi" punctuation   initial-quote)
  58     ("Pf" punctuation   final-quote)
  59     ("Po" punctuation   other)
  60     ("Sm" symbol        math)
  61     ("Sc" symbol        currency)
  62     ("Sk" symbol        modifier)
  63     ("So" symbol        other)
  64     ))
  65
  66 (defconst shuowen-radicals
  67   [?一 ?上 ?示 ?三 ?王 ?玉 ?玨 ?气 ?士 ?丨 ; 010
  68    ?屮 ?艸 ?蓐 ?茻 ?小 ?八 ?釆 ?半 ?牛 ?犛 ; 020
  69    ?告 ?口 ?凵 ?吅 ?哭 ?走 ?止 ?癶 ?步 ?此 ; 030
  70    ?正 ?是 ?辵 ?彳 ?廴 ?㢟 ?行 ?齒 ?牙 ?足 ; 040
  71    ?疋 ?品 ?龠 ?冊 ?㗊 ?舌 ?干 ?谷 ?只 ?㕯 ; 050
  72    ?句 ?丩 ?古 ?十 ?卅 ?言 ?誩 ?音 ?䇂 ?丵 ; 060
  73    ?菐 ?𠬞 ?𠬜 ?共 ?異 ?舁 ?𦥑 ?䢅 ?爨 ?革 ; 070
  74    ?鬲 ?䰜 ?爪 ?𠃨 ?鬥 ?又 ?𠂇 ?㕜 ?支 ?𦘒 ; 080
  75    ?聿 ?畫 ?隶 ?臤 ?臣 ?殳 ?殺 ?𠘧 ?寸 ?皮 ; 090
  76    ?㼱 ?攴 ?敎 ?卜 ?用 ?爻 ?㸚 ?𥄎 ?目 ?䀠 ; 100
  77    ?眉 ?盾 ?自 ?白 ?鼻 ?皕 ?習 ?羽 ?隹 ?奞 ; 110
  78    ?萑 ?𦫳 ?苜 ?羊 ?羴 ?瞿 ?雔 ?雥 ?鳥 ?烏 ; 120
  79    ?𠦒 ?冓 ?幺 ?𢆶 ?叀 ?玄 ?予 ?放 ?𠬪 ?𣦼 ; 130
  80    ?歺 ?死 ?冎 ?骨 ?肉 ?筋 ?刀 ?刃 ?㓞 ?丰 ; 140
  81    ?耒 ?𧢲 ?竹 ?箕 ?丌 ?左 ?工 ?㠭 ?巫 ?甘 ; 150
  82    ?曰 ?乃 ?丂 ?可 ?兮 ?号 ?亏 ?旨 ?喜 ?壴 ; 160
  83 ;  ?旨 ?曰 ?乃 ?丂 ?可 ?兮 ?号 ?亏 ?喜 ?壴 ; 160
  84    ?鼓 ?豈 ?豆 ?豊 ?豐 ?䖒 ?虍 ?虎 ?虤 ?皿 ; 170
  85    ?𠙴 ?去 ?血 ?丶 ?丹 ?青 ?井 ?皀 ?鬯 ?食 ; 180
  86    ?亼 ?會 ?倉 ?入 ?缶 ?矢 ?高 ?冂 ?𩫏 ?京 ; 190
  87    ?亯 ?𣆪 ?畗 ?㐭 ?嗇 ?來 ?麥 ?夊 ?舛 ?䑞 ; 200
  88    ?韋 ?弟 ?夂 ?久 ?桀 ?木 ?東 ?林 ?才 ?叒 ; 210
  89    ?之 ?帀 ?出 ?𣎵 ?生 ?乇 ?𠂹 ?𠌶 ?華 ?𥝌 ; 220
  90    ?稽 ?巢 ?桼 ?束 ?㯻 ?囗 ?員 ?貝 ?邑 ?𨛜 ; 230
  91    ?日 ?旦 ?倝 ?㫃 ?冥 ?晶 ?月 ?有 ?明 ?囧 ; 240
  92    ?夕 ?多 ?毌 ?𢎘 ?𣐺 ?卣 ?齊 ?朿 ?片 ?鼎 ; 250
  93    ?克 ?彔 ?禾 ?秝 ?黍 ?香 ?米 ?毇 ?臼 ?凶 ; 260
  94    ?𣎳 ?林 ?麻 ?尗 ?耑 ?韭 ?瓜 ?瓠 ?宀 ?宮 ; 270
  95    ?呂 ?穴 ?㝱 ?𤕫 ?冖 ?𠔼 ?冃 ?㒳 ?网 ?襾 ; 280
  96    ?巾 ?巿 ?帛 ?白 ?㡀 ?黹 ?人 ?𠤎 ?匕 ?从 ; 290
  97    ?比 ?北 ?丘 ?㐺 ?𡈼 ?重 ?臥 ?身 ?㐆 ?衣 ; 300
  98    ?裘 ?老 ?毛 ?毳 ?尸 ?尺 ?尾 ?履 ?舟 ?方 ; 310
  99    ?儿 ?兄 ?兂 ?皃 ?𠑹 ?先 ?秃 ?見 ?覞 ?欠 ; 320
 100    ?㱃 ?㳄 ?旡 ?頁 ?𦣻 ?面 ?丏 ?首 ?𥄉 ?須 ; 330
 101    ?彡 ?彣 ?文 ?髟 ?后 ?司 ?卮 ?卩 ?印 ?色 ; 340
 102    ?𠨍 ?辟 ?勹 ?包 ?茍 ?鬼 ?甶 ?厶 ?嵬 ?山 ; 350
 103    ?屾 ?屵 ?广 ?厂 ?丸 ?危 ?石 ?長 ?勿 ?冄 ; 360
 104    ?而 ?豕 ?㣇 ?彑 ?豚 ?豸 ?𤉡 ?易 ?象 ?馬 ; 370
 105    ?𢊁 ?鹿 ?麤 ?㲋 ?兔 ?萈 ?犬 ?㹜 ?鼠 ?能 ; 380
 106    ?熊 ?火 ?炎 ?黑 ?囪 ?焱 ?炙 ?赤 ?大 ?亦 ; 390
 107    ])
 108
 109 (defun shuowen-radical (number)
 110   (aref shuowen-radicals (1- number)))
 111
 112 (defvar char-db-file-coding-system 'utf-8-mcs-er)
 113
 114 (defvar char-db-ignored-attributes '(ideographic-products))
 115
 116 (defvar char-db-coded-charset-priority-list
 117   '(ascii
 118     control-1
 119     latin-iso8859-1
 120     latin-iso8859-2
 121     latin-iso8859-3
 122     latin-iso8859-4
 123     latin-iso8859-9
 124     latin-jisx0201
 125     cyrillic-iso8859-5
 126     greek-iso8859-7
 127     thai-tis620
 128     ;; =mj
 129     ;; =adobe-japan1-0
 130     ;; =adobe-japan1-1
 131     ;; =adobe-japan1-2
 132     ;; =adobe-japan1-3
 133     ;; =adobe-japan1-4
 134     ;; =adobe-japan1-5
 135     ;; =adobe-japan1-6
 136     =jis-x0208
 137     =jis-x0208@1978
 138     =jis-x0208@1983
 139     japanese-jisx0212
 140     chinese-gb2312
 141     =jis-x0208@1990
 142     chinese-cns11643-1
 143     chinese-cns11643-2
 144     chinese-cns11643-3
 145     chinese-cns11643-4
 146     chinese-cns11643-5
 147     chinese-cns11643-6
 148     chinese-cns11643-7
 149     =jis-x0213-1
 150     =jis-x0213-1@2000
 151     =jis-x0213-1@2004
 152     =jis-x0213-2
 153     korean-ksc5601
 154     chinese-isoir165
 155     katakana-jisx0201
 156     hebrew-iso8859-8
 157     chinese-gb12345
 158     latin-viscii
 159     ethiopic-ucs
 160     =big5-cdp
 161     =hanyo-denshi/ja
 162     =hanyo-denshi/jb
 163     =hanyo-denshi/jc
 164     =hanyo-denshi/jd
 165     =hanyo-denshi/ft
 166     =hanyo-denshi/ia
 167     =hanyo-denshi/ib
 168     =hanyo-denshi/hg
 169     =hanyo-denshi/jt
 170     =hanyo-denshi/ks
 171     =hanyo-denshi/tk
 172     ;; ==mj
 173     ;; ==adobe-japan1-0
 174     ;; ==adobe-japan1-1
 175     ;; ==adobe-japan1-2
 176     ;; ==adobe-japan1-3
 177     ;; ==adobe-japan1-4
 178     ;; ==adobe-japan1-5
 179     ;; ==adobe-japan1-6
 180     ==jis-x0208
 181     ==jis-x0213-1
 182     ==jis-x0213-2
 183     ==jis-x0212
 184     ==hanyo-denshi/ja
 185     ==hanyo-denshi/jb
 186     ==hanyo-denshi/jc
 187     ==hanyo-denshi/ft
 188     ==hanyo-denshi/ia
 189     ==hanyo-denshi/ib
 190     ==hanyo-denshi/hg
 191     ==hanyo-denshi/jt
 192     ==hanyo-denshi/ks
 193     ==hanyo-denshi/tk
 194     =gt
 195     =gt-k
 196     =daikanwa
 197     =daikanwa@rev2
 198     =daikanwa@rev1
 199     =daikanwa/+p
 200     ==daikanwa
 201     =cbeta
 202     ideograph-hanziku-1
 203     ideograph-hanziku-2
 204     ideograph-hanziku-3
 205     ideograph-hanziku-4
 206     ideograph-hanziku-5
 207     ideograph-hanziku-6
 208     ideograph-hanziku-7
 209     ideograph-hanziku-8
 210     ideograph-hanziku-9
 211     ideograph-hanziku-10
 212     ideograph-hanziku-11
 213     ideograph-hanziku-12
 214     ;; =>>>adobe-japan1-0
 215     ;; =>>>adobe-japan1-1
 216     ;; =>>>adobe-japan1-2
 217     ;; =>>>adobe-japan1-3
 218     ;; =>>>adobe-japan1-4
 219     ;; =>>>adobe-japan1-5
 220     ;; =>>>adobe-japan1-6
 221     ;; =>>>jis-x0208
 222     ;; =>>>jis-x0213-1
 223     ;; =>>>jis-x0213-2
 224     ;; =>>>hanyo-denshi/ja
 225     ;; =>>>hanyo-denshi/jb
 226     ;; =>>>hanyo-denshi/jc
 227     ;; =>>>hanyo-denshi/ft
 228     ;; =>>>hanyo-denshi/ib
 229     ;; =>>>hanyo-denshi/hg
 230     ;; =>>>hanyo-denshi/jt
 231     ;; =>>>hanyo-denshi/ks
 232     ;; =>>>gt
 233     =>>adobe-japan1-0
 234     =>>adobe-japan1-1
 235     =>>adobe-japan1-2
 236     =>>adobe-japan1-3
 237     =>>adobe-japan1-4
 238     =>>adobe-japan1-5
 239     =>>adobe-japan1-6
 240     =>>jis-x0208
 241     =>>jis-x0213-1
 242     =>>jis-x0213-1@2000
 243     =>>jis-x0213-1@2004
 244     =>>jis-x0213-2
 245     =>>jis-x0208@1978
 246     =>>hanyo-denshi/ft
 247     =>>hanyo-denshi/jt
 248     =>>hanyo-denshi/ks
 249     =>>gt
 250     =>>daikanwa
 251     =>>cbeta
 252     =+>jis-x0208
 253     =+>jis-x0213-1
 254     =+>jis-x0213-2
 255     =+>adobe-japan1-0
 256     =+>adobe-japan1-1
 257     =+>adobe-japan1-2
 258     =+>adobe-japan1-3
 259     =+>adobe-japan1-4
 260     =+>adobe-japan1-5
 261     =+>adobe-japan1-6
 262     =+>jis-x0208@1978
 263     =>jis-x0208
 264     =>jis-x0208@1997
 265     =>jis-x0213-1
 266     =>jis-x0213-1@2000
 267     =>jis-x0213-1@2004
 268     =>jis-x0213-2
 269     ==>ucs@bucs
 270     =>iwds-1
 271     ;; =>ucs@hanyo-denshi
 272     =>ucs@iso
 273     =>ucs@unicode
 274     =>ucs@jis
 275     =>ucs@cns
 276     =>ucs@ks
 277     =+>ucs@iso
 278     =+>ucs@unicode
 279     =+>ucs@jis
 280     =+>ucs@jis/1990
 281     =+>ucs@cns
 282     =+>ucs@ks
 283     =>>ucs@iso
 284     =>>ucs@unicode
 285     =>>ucs@jis
 286     =>>ucs@cns
 287     =>>>ucs@iso
 288     =>>>ucs@unicode
 289     ==ucs@iso
 290     ==ucs@unicode
 291     ;; ==ucs@cns
 292     ==gb2312
 293     ==ks-x1001
 294     ==cns11643-1
 295     ==cns11643-2
 296     ==cns11643-3
 297     ==cns11643-4
 298     ==cns11643-5
 299     ==cns11643-6
 300     ==cns11643-7
 301     ==gt
 302     ==jis-x0208@1990
 303     ;; ==jis-x0208@1983
 304     ==jis-x0208@1978
 305     ==gt-k
 306     =ucs@iso
 307     =ucs@unicode
 308     =ucs@cns
 309     ==big5-cdp
 310     ==cbeta
 311     =>>big5-cdp
 312     =>>gt-k
 313     =+>gt
 314     =>gt
 315     =>mj
 316     =>big5-cdp
 317     =>daikanwa
 318     =>daikanwa/ho
 319     =>cns11643-5
 320     =>cns11643-7
 321     =big5
 322     =big5-eten
 323     =>gt-k
 324     =zinbun-oracle
 325     =>zinbun-oracle
 326     =ruimoku-v6
 327     =>>ruimoku-v6
 328     ==ruimoku-v6
 329     =jef-china3
 330     =>cbeta
 331     =shinjigen
 332     =ucs-var-001
 333     =ucs-var-002
 334     =ucs-var-003
 335     =ucs-var-004
 336     =ucs-var-005
 337     =ucs-var-006
 338     =ucs-var-008
 339     =ucs-var-010
 340     =ucs-itaiji-001
 341     =ucs-itaiji-002
 342     =ucs-itaiji-003
 343     =ucs-itaiji-004
 344     =ucs-itaiji-005
 345     =ucs-itaiji-006
 346     =ucs-itaiji-007
 347     =ucs-itaiji-008
 348     =ucs-itaiji-009
 349     =ucs-itaiji-010
 350     =ucs-itaiji-011
 351     =>ucs-itaiji-001
 352     =>ucs-itaiji-002
 353     =>ucs-itaiji-003
 354     =>ucs-itaiji-004
 355     =>ucs-itaiji-005
 356     =>ucs-itaiji-006
 357     =>ucs-itaiji-007
 358     =>ucs-itaiji-009
 359     =big5-cdp-var-001
 360     =big5-cdp-var-002
 361     =big5-cdp-var-003
 362     =big5-cdp-var-004
 363     =big5-cdp-var-005
 364     =big5-cdp-itaiji-001
 365     =big5-cdp-itaiji-002
 366     =big5-cdp-itaiji-003
 367     =>big5-cdp-itaiji-001
 368     =>ucs@iwds-1
 369     =>ucs@cognate
 370     =>ucs@component
 371     =>ucs-itaiji-001@iwds-1
 372     ==ucs@gb
 373     =ucs@gb
 374     ==ucs-var-002
 375     =ucs@JP/hanazono))
 376
 377
 378 ;;; @ char-db formatters
 379 ;;;
 380
 381 (defun char-db-make-char-spec (char)
 382   (let (ret char-spec)
 383     (cond ((characterp char)
 384            (cond ((and (setq ret (encode-char char '=ucs 'defined-only))
 385                        (not (and (<= #xE000 ret)(<= ret #xF8FF))))
 386                   (setq char-spec (list (cons '=ucs ret)))
 387                   (cond ((setq ret (get-char-attribute char 'name))
 388                          (setq char-spec (cons (cons 'name ret) char-spec))
 389                          )
 390                         ((setq ret (get-char-attribute char 'name*))
 391                          (setq char-spec (cons (cons 'name* ret) char-spec))
 392                          ))
 393                   )
 394                  ((encode-char char '=mj 'defined-only)
 395                   (setq char-spec nil)
 396                   (dolist (ccs (charset-list))
 397                     (if (and (or (eq ccs '=mj)
 398                                  ;; (eq (charset-property ccs 'iso-ir) 177)
 399                                  (string-match "=ucs@" (symbol-name ccs))
 400                                  )
 401                              (setq ccs (charset-name ccs))
 402                              (null (assq ccs char-spec))
 403                              (setq ret (encode-char char ccs 'defined-only)))
 404                         (setq char-spec (cons (cons ccs ret) char-spec))))
 405                   )
 406                  ((encode-char char '==mj 'defined-only)
 407                   (setq char-spec nil)
 408                   (dolist (ccs (charset-list))
 409                     (if (and (or (eq ccs '==mj)
 410                                  ;; (eq (charset-property ccs 'iso-ir) 177)
 411                                  (string-match "=ucs@" (symbol-name ccs))
 412                                  )
 413                              (setq ccs (charset-name ccs))
 414                              (null (assq ccs char-spec))
 415                              (setq ret (encode-char char ccs 'defined-only)))
 416                         (setq char-spec (cons (cons ccs ret) char-spec))))
 417                   )
 418                  ((encode-char char '=adobe-japan1 'defined-only)
 419                   (setq char-spec nil)
 420                   (dolist (ccs (charset-list))
 421                     (if (and (or (memq ccs
 422                                        '(=adobe-japan1-0
 423                                          =adobe-japan1-1
 424                                          =adobe-japan1-2
 425                                          =adobe-japan1-3
 426                                          =adobe-japan1-4
 427                                          =adobe-japan1-5
 428                                          =adobe-japan1-6
 429                                          ))
 430                                  ;; (eq (charset-property ccs 'iso-ir) 177)
 431                                  (string-match "=ucs@" (symbol-name ccs))
 432                                  )
 433                              (setq ccs (charset-name ccs))
 434                              (null (assq ccs char-spec))
 435                              (setq ret (encode-char char ccs 'defined-only)))
 436                         (setq char-spec (cons (cons ccs ret) char-spec))))
 437                   )
 438                  ((encode-char char '==adobe-japan1 'defined-only)
 439                   (setq char-spec nil)
 440                   (dolist (ccs (charset-list))
 441                     (if (and (or (memq ccs
 442                                        '(==adobe-japan1-0
 443                                          ==adobe-japan1-1
 444                                          ==adobe-japan1-2
 445                                          ==adobe-japan1-3
 446                                          ==adobe-japan1-4
 447                                          ==adobe-japan1-5
 448                                          ==adobe-japan1-6
 449                                          ))
 450                                  ;; (eq (charset-property ccs 'iso-ir) 177)
 451                                  (string-match "=ucs@" (symbol-name ccs))
 452                                  )
 453                              (setq ccs (charset-name ccs))
 454                              (null (assq ccs char-spec))
 455                              (setq ret (encode-char char ccs 'defined-only)))
 456                         (setq char-spec (cons (cons ccs ret) char-spec))))
 457                   )
 458                  ((setq ret
 459                         (catch 'tag
 460                           (let ((rest char-db-coded-charset-priority-list)
 461                                 ccs)
 462                             (while rest
 463                               (setq ccs (charset-name
 464                                          (find-charset (car rest))))
 465                               (if (setq ret
 466                                         (encode-char char ccs
 467                                                      'defined-only))
 468                                   (throw 'tag (cons ccs ret)))
 469                               (setq rest (cdr rest))))))
 470                   (setq char-spec (list ret))
 471                   (dolist (ccs (delq (car ret) (charset-list)))
 472                     (if (and (or (charset-iso-final-char ccs)
 473                                  (memq ccs
 474                                        '(=daikanwa
 475                                          =daikanwa@rev2
 476                                          ;; =gt-k
 477                                          =jis-x0208@1997
 478                                          ))
 479                                  (and (eq (charset-property ccs 'iso-ir) 177)
 480                                       (not (eq (charset-name ccs) '=ucs@big5)))
 481                                  ;; (string-match "=ucs@" (symbol-name ccs))
 482                                  )
 483                              (setq ccs (charset-name ccs))
 484                              (null (assq ccs char-spec))
 485                              (setq ret (encode-char char ccs 'defined-only)))
 486                         (setq char-spec (cons (cons ccs ret) char-spec))))
 487                   (if (null char-spec)
 488                       (setq char-spec (split-char char)))
 489                   (cond ((setq ret (get-char-attribute char 'name))
 490                          (setq char-spec (cons (cons 'name ret) char-spec))
 491                          )
 492                         ((setq ret (get-char-attribute char 'name*))
 493                          (setq char-spec (cons (cons 'name* ret) char-spec))
 494                          ))
 495                   )
 496                  ((setq ret (get-char-attribute
 497                              char 'ideographic-combination))
 498                   (setq char-spec
 499                         (cons (cons 'ideographic-combination ret)
 500                               char-spec))
 501                   ))
 502            char-spec)
 503           ((consp char)
 504            char))))
 505
 506 (defun char-db-insert-char-spec (char &optional readable column
 507                                       required-features)
 508   (unless column
 509     (setq column (current-column)))
 510   (let (char-spec temp-char)
 511     (setq char-spec (char-db-make-char-spec char))
 512     (unless (or (characterp char) ; char
 513                 (condition-case nil
 514                     (setq char (find-char char-spec))
 515                   (error nil)))
 516       ;; define temporary character
 517       ;;   Current implementation is dirty.
 518       (setq temp-char (define-char (cons '(ideograph-daikanwa . 0)
 519                                          char-spec)))
 520       (remove-char-attribute temp-char 'ideograph-daikanwa)
 521       (setq char temp-char))
 522     (insert-char-attributes char
 523                             readable
 524                             (union (mapcar #'car char-spec)
 525                                    required-features)
 526                             nil 'for-sub-node)
 527     (when temp-char
 528       ;; undefine temporary character
 529       ;;   Current implementation is dirty.
 530       (setq char-spec (char-attribute-alist temp-char))
 531       (while char-spec
 532         (remove-char-attribute temp-char (car (car char-spec)))
 533         (setq char-spec (cdr char-spec))))))
 534
 535 (defun char-db-insert-alist (alist &optional readable column)
 536   (unless column
 537     (setq column (current-column)))
 538   (let ((line-breaking
 539          (concat "\n" (make-string (1+ column) ?\ )))
 540         name value
 541         ret al ; cal
 542         key
 543         lbs cell rest separator)
 544     (insert "(")
 545     (while alist
 546       (setq name (car (car alist))
 547             value (cdr (car alist)))
 548       (cond ((eq name 'char)
 549              (insert "(char . ")
 550              (if (setq ret (condition-case nil
 551                                (find-char value)
 552                              (error nil)))
 553                  (progn
 554                    (setq al nil
 555                          ;; cal nil
 556                          )
 557                    (while value
 558                      (setq key (car (car value)))
 559                      ;; (if (find-charset key)
 560                      ;;     (setq cal (cons key cal))
 561                      (setq al (cons key al))
 562                      ;; )
 563                      (setq value (cdr value)))
 564                    (insert-char-attributes ret
 565                                            readable
 566                                            (or al 'none) ; cal
 567                                            nil 'for-sub-node))
 568                (insert (prin1-to-string value)))
 569              (insert ")")
 570              (insert line-breaking))
 571             ((consp value)
 572              (insert (format "(%-18s " name))
 573              (setq lbs (concat "\n" (make-string (current-column) ?\ )))
 574              (while (consp value)
 575                (setq cell (car value))
 576                (if (and (consp cell)
 577                         (consp (car cell))
 578                         (setq ret (condition-case nil
 579                                       (find-char cell)
 580                                     (error nil)))
 581                         )
 582                    (progn
 583                      (setq rest cell
 584                            al nil
 585                            ;; cal nil
 586                            )
 587                      (while rest
 588                        (setq key (car (car rest)))
 589                        ;; (if (find-charset key)
 590                        ;;     (setq cal (cons key cal))
 591                        (setq al (cons key al))
 592                        ;; )
 593                        (setq rest (cdr rest)))
 594                      (if separator
 595                          (insert lbs))
 596                      (insert-char-attributes ret
 597                                              readable
 598                                              al ; cal
 599                                              nil 'for-sub-node)
 600                      (setq separator lbs))
 601                  (if separator
 602                      (insert separator))
 603                  (insert (prin1-to-string cell))
 604                  (setq separator " "))
 605                (setq value (cdr value)))
 606              (insert ")")
 607              (insert line-breaking))
 608             (t
 609              (insert (format "(%-18s . %S)%s"
 610                              name value
 611                              line-breaking))))
 612       (setq alist (cdr alist))))
 613   (insert ")"))
 614
 615 (defun char-db-insert-char-reference (plist &optional readable column)
 616   (unless column
 617     (setq column (current-column)))
 618   (let ((line-breaking
 619          (concat "\n" (make-string (1+ column) ?\ )))
 620         (separator "")
 621         name value)
 622     (insert "(")
 623     (while plist
 624       (setq name (pop plist))
 625       (setq value (pop plist))
 626       (cond ((eq name :char)
 627              (insert separator)
 628              (insert ":char\t")
 629              (cond ((numberp value)
 630                     (setq value (decode-char '=ucs value)))
 631                    ;; ((consp value)
 632                    ;;  (setq value (or (find-char value)
 633                    ;;                  value)))
 634                    )
 635              (char-db-insert-char-spec value readable)
 636              (insert line-breaking)
 637              (setq separator ""))
 638             ((eq name :radical)
 639              (insert (format "%s%s\t%d ; %c%s"
 640                              separator
 641                              name value
 642                              (ideographic-radical value)
 643                              line-breaking))
 644              (setq separator ""))
 645             (t
 646              (insert (format "%s%s\t%S" separator name value))
 647              (setq separator line-breaking)))
 648       ))
 649   (insert ")"))
 650
 651 (defun char-db-decode-isolated-char (ccs code-point)
 652   (let (ret)
 653     (setq ret
 654           (cond ((eq ccs 'arabic-iso8859-6)
 655                  (decode-char ccs code-point))
 656                 ;; ((eq ccs '=gt)
 657                 ;;  (decode-builtin-char '==gt code-point))
 658                 ((and (memq ccs '(=gt-pj-1
 659                                   =gt-pj-2
 660                                   =gt-pj-3
 661                                   =gt-pj-4
 662                                   =gt-pj-5
 663                                   =gt-pj-6
 664                                   =gt-pj-7
 665                                   =gt-pj-8
 666                                   =gt-pj-9
 667                                   =gt-pj-10
 668                                   =gt-pj-11))
 669                       (setq ret (decode-char ccs code-point))
 670                       (setq ret (encode-char ret '=gt 'defined-only)))
 671                  (decode-builtin-char '=gt ret))
 672                 (t
 673                  (decode-builtin-char ccs code-point))))
 674     (cond ((null ret)
 675            (or (decode-char ccs code-point)
 676                (define-char (list (cons ccs code-point)))))
 677           ((and (<= 0 (char-int ret))
 678                 (<= (char-int ret) #x1F))
 679            (decode-char '=ucs (+ #x2400 (char-int ret))))
 680           ((= (char-int ret) #x7F)
 681            ?\u2421)
 682           (t ret))))
 683
 684 (defvar char-db-convert-obsolete-format t)
 685
 686 (defun char-db-insert-ccs-feature (name value line-breaking)
 687   (cond
 688    ((integerp value)
 689     (insert
 690      (format
 691       (cond
 692        ((memq name '(=>iwds-1
 693                      ==shinjigen
 694                      =shinjigen
 695                      =shinjigen@1ed ==shinjigen@1ed
 696                      =shinjigen@rev ==shinjigen@rev
 697                      =shinjigen/+p@rev ==shinjigen/+p@rev
 698                      ===daikanwa/ho ==daikanwa/ho
 699                      =daikanwa/ho =>>daikanwa/ho =>daikanwa/ho))
 700         "(%-18s .  %04d)\t; %c")
 701        ((eq name '=shinjigen@1ed/24pr)
 702         "(%-18s . %04d)\t; %c")
 703        ((or
 704          (memq name
 705                '(===daikanwa
 706                  ==daikanwa =daikanwa =>>daikanwa =>daikanwa
 707                  =daikanwa@rev1 =daikanwa@rev2
 708                  =daikanwa/+p ==daikanwa/+p ===daikanwa/+p
 709                  =>>daikanwa/+p
 710                  =daikanwa/+2p =>>daikanwa/+2p
 711                  =gt ==gt ===gt
 712                  =>>gt =+>gt =>gt
 713                  =gt-k ==gt-k ===gt-k
 714                  =>>gt-k =>gt-k
 715                  =adobe-japan1-0 ==adobe-japan1-0 ===adobe-japan1-0
 716                  =adobe-japan1-1 ==adobe-japan1-1 ===adobe-japan1-1
 717                  =adobe-japan1-2 ==adobe-japan1-2 ===adobe-japan1-2
 718                  =adobe-japan1-3 ==adobe-japan1-3 ===adobe-japan1-3
 719                  =adobe-japan1-4 ==adobe-japan1-4 ===adobe-japan1-4
 720                  =adobe-japan1-5 ==adobe-japan1-5 ===adobe-japan1-5
 721                  =adobe-japan1-6 ==adobe-japan1-6 ===adobe-japan1-6
 722                  =>>adobe-japan1-0 =+>adobe-japan1-0
 723                  =>>adobe-japan1-1 =+>adobe-japan1-1
 724                  =>>adobe-japan1-2 =+>adobe-japan1-2
 725                  =>>adobe-japan1-3 =+>adobe-japan1-3
 726                  =>>adobe-japan1-4 =+>adobe-japan1-4
 727                  =>>adobe-japan1-5 =+>adobe-japan1-5
 728                  =>>adobe-japan1-6 =+>adobe-japan1-6
 729                  =>cbeta =cbeta =>>cbeta ==cbeta ===cbeta
 730                  =zinbun-oracle =>zinbun-oracle
 731                  ===hng-jou ===hng-keg ===hng-dng ===hng-mam
 732                  ===hng-drt ===hng-kgk ===hng-myz ===hng-kda
 733                  ===hng-khi ===hng-khm ===hng-hok ===hng-kyd ===hng-sok
 734                  ===hng-yhk ===hng-kak ===hng-kar ===hng-kae
 735                  ===hng-sys ===hng-tsu ===hng-tzj
 736                  ===hng-hos ===hng-nak ===hng-jhk
 737                  ===hng-hod ===hng-gok ===hng-ink ===hng-nto
 738                  ===hng-nkm ===hng-k24 ===hng-nkk
 739                  ===hng-kcc ===hng-kcj ===hng-kbk ===hng-sik
 740                  ===hng-skk ===hng-kyu ===hng-ksk ===hng-wan
 741                  ===hng-okd ===hng-wad ===hng-kmi
 742                  ===hng-zkd ===hng-doh ===hng-jyu
 743                  ===hng-tzs ===hng-kss ===hng-kyo
 744                  ===hng-smk))
 745          ;; (string-match "^=adobe-" (symbol-name name))
 746          )
 747         "(%-18s . %05d)\t; %c")
 748        ((memq name '(=hanyo-denshi/ks
 749                      ==hanyo-denshi/ks ===hanyo-denshi/ks
 750                      =>>hanyo-denshi/ks
 751                      =koseki ==koseki
 752                      =mj ==mj ===mj =>>mj =>mj
 753                      =zihai mojikyo))
 754         "(%-18s . %06d)\t; %c")
 755        ((memq name '(=hanyo-denshi/tk ==hanyo-denshi/tk))
 756         "(%-18s . %08d)\t; %c")
 757        ((>= (charset-dimension name) 2)
 758         "(%-18s . #x%04X)\t; %c")
 759        (t
 760         "(%-18s . #x%02X)\t; %c"))
 761       name
 762       (if (= (charset-iso-graphic-plane name) 1)
 763           (logior value
 764                   (cond ((= (charset-dimension name) 1)
 765                          #x80)
 766                         ((= (charset-dimension name) 2)
 767                          #x8080)
 768                         ((= (charset-dimension name) 3)
 769                          #x808080)
 770                         (t 0)))
 771         value)
 772       (char-db-decode-isolated-char name value)))
 773     (if (and (= (charset-chars name) 94)
 774              (= (charset-dimension name) 2))
 775         (insert (format " [%02d-%02d]"
 776                         (- (lsh value -8) 32)
 777                         (- (logand value 255) 32))))
 778     )
 779    (t
 780     (insert (format "(%-18s . %s)" name value))
 781     ))
 782   (insert line-breaking))
 783
 784 (defun char-db-insert-relation-feature (char name value line-breaking
 785                                              ccss readable)
 786   (insert (format "(%-18s%s " name line-breaking))
 787   (let ((lbs (concat "\n" (make-string (current-column) ?\ )))
 788         separator cell sources required-features
 789         ret)
 790     (while (consp value)
 791       (setq cell (car value))
 792       (if (integerp cell)
 793           (setq cell (decode-char '=ucs cell)))
 794       (cond
 795        ((eq name '->subsumptive)
 796         (when (or (not (some (lambda (atr)
 797                                (get-char-attribute cell atr))
 798                              char-db-ignored-attributes))
 799                   (some (lambda (ccs)
 800                           (encode-char cell ccs 'defined-only))
 801                         ccss))
 802           (if separator
 803               (insert lbs))
 804           (let ((char-db-ignored-attributes
 805                  (cons '<-subsumptive
 806                        char-db-ignored-attributes)))
 807             (insert-char-attributes cell readable nil nil 'for-sub-node))
 808           (setq separator lbs))
 809         )
 810        ((characterp cell)
 811         (setq sources
 812               (get-char-attribute
 813                char (intern (format "%s*sources" name))))
 814         (setq required-features nil)
 815         (dolist (source sources)
 816           (cond
 817            ((memq source '(JP
 818                            JP/Jouyou
 819                            shinjigen shinjigen@1ed shinjigen@rev))
 820             (setq required-features
 821                   (union required-features
 822                          '(=jis-x0208
 823                            =jis-x0208@1990
 824                            =jis-x0213-1@2000
 825                            =jis-x0213-1@2004
 826                            =jis-x0213-2
 827                            =jis-x0212
 828                            =jis-x0208@1983
 829                            =jis-x0208@1978
 830                            =shinjigen
 831                            =shinjigen@1ed
 832                            =shinjigen@rev
 833                            =shinjigen/+p@rev))))
 834            ((eq source 'CN)
 835             (setq required-features
 836                   (union required-features
 837                          '(=gb2312
 838                            =gb12345
 839                            =iso-ir165)))))
 840           (cond
 841            ((find-charset (setq ret (intern (format "=%s" source))))
 842             (setq required-features
 843                   (cons ret required-features)))
 844            (t (setq required-features
 845                     (cons source required-features)))))
 846         (cond ((string-match "@JP" (symbol-name name))
 847                (setq required-features
 848                      (union required-features
 849                             '(=jis-x0208
 850                               =jis-x0208@1990
 851                               =jis-x0213-1-2000
 852                               =jis-x0213-2-2000
 853                               =jis-x0212
 854                               =jis-x0208@1983
 855                               =jis-x0208@1978))))
 856               ((string-match "@CN" (symbol-name name))
 857                (setq required-features
 858                      (union required-features
 859                             '(=gb2312
 860                               =gb12345
 861                               =iso-ir165)))))
 862         (if separator
 863             (insert lbs))
 864         (if readable
 865             (insert (format "%S" cell))
 866           (char-db-insert-char-spec cell readable
 867                                     nil
 868                                     required-features))
 869         (setq separator lbs))
 870        ((consp cell)
 871         (if separator
 872             (insert lbs))
 873         (if (consp (car cell))
 874             (char-db-insert-char-spec cell readable)
 875           (char-db-insert-char-reference cell readable))
 876         (setq separator lbs))
 877        (t
 878         (if separator
 879             (insert separator))
 880         (insert (prin1-to-string cell))
 881         (setq separator " ")))
 882       (setq value (cdr value)))
 883     (insert ")")
 884     (insert line-breaking)))
 885
 886 (defun insert-char-attributes (char &optional readable attributes column
 887                                     for-sub-node)
 888   (unless column
 889     (setq column (current-column)))
 890   (let (name value ; has-long-ccs-name
 891         rest
 892         radical strokes
 893         (line-breaking
 894          (concat "\n" (make-string (1+ column) ?\ )))
 895         lbs cell separator ret
 896         key al cal
 897         dest-ccss ; sources required-features
 898         ccss)
 899     (let (atr-d)
 900       (setq attributes
 901             (sort (if attributes
 902                       (if (consp attributes)
 903                           (progn
 904                             (dolist (name attributes)
 905                               (unless (memq name char-db-ignored-attributes)
 906                                 (if (find-charset name)
 907                                     (push name ccss))
 908                                 (push name atr-d)))
 909                             atr-d))
 910                     (dolist (name (char-attribute-list))
 911                       (unless (memq name char-db-ignored-attributes)
 912                         (if (find-charset name)
 913                             (push name ccss))
 914                         (push name atr-d)))
 915                     atr-d)
 916                   #'char-attribute-name<)))
 917     (insert "(")
 918     (when (memq '<-subsumptive attributes)
 919       (when (or readable (not for-sub-node))
 920         (when (setq value (get-char-attribute char '<-subsumptive))
 921           (char-db-insert-relation-feature char '<-subsumptive value
 922                                            line-breaking
 923                                            ccss readable)))
 924       (setq attributes (delq '<-subsumptive attributes)))
 925     (when (and (memq '<-denotational attributes)
 926                (setq value (get-char-attribute char '<-denotational)))
 927       (char-db-insert-relation-feature char '<-denotational value
 928                                        line-breaking
 929                                        ccss readable)
 930       (setq attributes (delq '<-denotational attributes)))
 931     (when (and (memq '<-denotational@component attributes)
 932                (setq value (get-char-attribute char '<-denotational@component)))
 933       (char-db-insert-relation-feature char '<-denotational@component value
 934                                        line-breaking
 935                                        ccss readable)
 936       (setq attributes (delq '<-denotational@component attributes)))
 937     (when (and (memq 'name attributes)
 938                (setq value (get-char-attribute char 'name)))
 939       (insert (format
 940                (if (> (+ (current-column) (length value)) 48)
 941                    "(name . %S)%s"
 942                  "(name               . %S)%s")
 943                value line-breaking))
 944       (setq attributes (delq 'name attributes))
 945       )
 946     (when (and (memq 'name* attributes)
 947                (setq value (get-char-attribute char 'name*)))
 948       (insert (format
 949                (if (> (+ (current-column) (length value)) 48)
 950                    "(name* . %S)%s"
 951                  "(name*              . %S)%s")
 952                value line-breaking))
 953       (setq attributes (delq 'name* attributes))
 954       )
 955     (when (and (memq 'script attributes)
 956                (setq value (get-char-attribute char 'script)))
 957       (insert (format "(script\t\t%s)%s"
 958                       (mapconcat (function prin1-to-string)
 959                                  value " ")
 960                       line-breaking))
 961       (setq attributes (delq 'script attributes))
 962       )
 963     (dolist (name '(=>ucs =>ucs*))
 964       (when (and (memq name attributes)
 965                  (setq value (get-char-attribute char name)))
 966         (insert (format "(%-18s . #x%04X)\t; %c%s"
 967                         name value (decode-char '=ucs value)
 968                         line-breaking))
 969         (setq attributes (delq name attributes))))
 970     (dolist (name '(=>ucs@gb =>ucs@big5))
 971       (when (and (memq name attributes)
 972                  (setq value (get-char-attribute char name)))
 973         (insert (format "(%-18s . #x%04X)\t; %c%s"
 974                         name value
 975                         (decode-char (intern
 976                                       (concat "="
 977                                               (substring
 978                                                (symbol-name name) 2)))
 979                                      value)
 980                         line-breaking))
 981         (setq attributes (delq name attributes))
 982         ))
 983     ;; (dolist (name '(=>daikanwa))
 984     ;;   (when (and (memq name attributes)
 985     ;;              (setq value (get-char-attribute char name)))
 986     ;;     (insert
 987     ;;      (if (integerp value)
 988     ;;          (format "(%-18s . %05d)\t; %c%s"
 989     ;;                  name value (decode-char '=daikanwa value)
 990     ;;                  line-breaking)
 991     ;;        (format "(%-18s %s)\t; %c%s"
 992     ;;                name
 993     ;;                (mapconcat (function prin1-to-string)
 994     ;;                           value " ")
 995     ;;                (char-representative-of-daikanwa char)
 996     ;;                line-breaking)))
 997     ;;     (setq attributes (delq name attributes))))
 998     (when (and (memq 'general-category attributes)
 999                (setq value (get-char-attribute char 'general-category)))
1000       (insert (format
1001                "(general-category\t%s) ; %s%s"
1002                (mapconcat (lambda (cell)
1003                             (format "%S" cell))
1004                           value " ")
1005                (cond ((rassoc value unidata-normative-category-alist)
1006                       "Normative Category")
1007                      ((rassoc value unidata-informative-category-alist)
1008                       "Informative Category")
1009                      (t
1010                       "Unknown Category"))
1011                line-breaking))
1012       (setq attributes (delq 'general-category attributes))
1013       )
1014     (when (and (memq 'bidi-category attributes)
1015                (setq value (get-char-attribute char 'bidi-category)))
1016       (insert (format "(bidi-category\t. %S)%s"
1017                       value
1018                       line-breaking))
1019       (setq attributes (delq 'bidi-category attributes))
1020       )
1021     (unless (or (not (memq 'mirrored attributes))
1022                 (eq (setq value (get-char-attribute char 'mirrored 'empty))
1023                     'empty))
1024       (insert (format "(mirrored\t\t. %S)%s"
1025                       value
1026                       line-breaking))
1027       (setq attributes (delq 'mirrored attributes))
1028       )
1029     (cond
1030      ((and (memq 'decimal-digit-value attributes)
1031            (setq value (get-char-attribute char 'decimal-digit-value)))
1032       (insert (format "(decimal-digit-value . %S)%s"
1033                       value
1034                       line-breaking))
1035       (setq attributes (delq 'decimal-digit-value attributes))
1036       (when (and (memq 'digit-value attributes)
1037                  (setq value (get-char-attribute char 'digit-value)))
1038         (insert (format "(digit-value\t . %S)%s"
1039                         value
1040                         line-breaking))
1041         (setq attributes (delq 'digit-value attributes))
1042         )
1043       (when (and (memq 'numeric-value attributes)
1044                  (setq value (get-char-attribute char 'numeric-value)))
1045         (insert (format "(numeric-value\t . %S)%s"
1046                         value
1047                         line-breaking))
1048         (setq attributes (delq 'numeric-value attributes))
1049         )
1050       )
1051      (t
1052       (when (and (memq 'digit-value attributes)
1053                  (setq value (get-char-attribute char 'digit-value)))
1054         (insert (format "(digit-value\t. %S)%s"
1055                         value
1056                         line-breaking))
1057         (setq attributes (delq 'digit-value attributes))
1058         )
1059       (when (and (memq 'numeric-value attributes)
1060                  (setq value (get-char-attribute char 'numeric-value)))
1061         (insert (format "(numeric-value\t. %S)%s"
1062                         value
1063                         line-breaking))
1064         (setq attributes (delq 'numeric-value attributes))
1065         )))
1066     (when (and (memq 'iso-10646-comment attributes)
1067                (setq value (get-char-attribute char 'iso-10646-comment)))
1068       (insert (format "(iso-10646-comment\t. %S)%s"
1069                       value
1070                       line-breaking))
1071       (setq attributes (delq 'iso-10646-comment attributes))
1072       )
1073     (when (and (memq 'morohashi-daikanwa attributes)
1074                (setq value (get-char-attribute char 'morohashi-daikanwa)))
1075       (insert (format "(morohashi-daikanwa\t%s)%s"
1076                       (mapconcat (function prin1-to-string) value " ")
1077                       line-breaking))
1078       (setq attributes (delq 'morohashi-daikanwa attributes))
1079       )
1080     (setq radical nil
1081           strokes nil)
1082     (when (and (memq 'ideographic-radical attributes)
1083                (setq value (get-char-attribute char 'ideographic-radical)))
1084       (setq radical value)
1085       (insert (format "(ideographic-radical . %S)\t; %c%s"
1086                       radical
1087                       (ideographic-radical radical)
1088                       line-breaking))
1089       (setq attributes (delq 'ideographic-radical attributes))
1090       )
1091     (when (and (memq 'shuowen-radical attributes)
1092                (setq value (get-char-attribute char 'shuowen-radical)))
1093       (insert (format "(shuowen-radical\t. %S)\t; %c%s"
1094                       value
1095                       (shuowen-radical value)
1096                       line-breaking))
1097       (setq attributes (delq 'shuowen-radical attributes))
1098       )
1099     (let (key)
1100       (dolist (domain
1101                (append
1102                 char-db-feature-domains
1103                 (let (dest domain)
1104                   (dolist (feature (char-attribute-list))
1105                     (setq feature (symbol-name feature))
1106                     (when (string-match
1107                            "\\(radical\\|strokes\\)@\\([^@*]+\\)\\(\\*\\|$\\)"
1108                            feature)
1109                       (setq domain (intern (match-string 2 feature)))
1110                      (unless (memq domain dest)
1111                        (setq dest (cons domain dest)))))
1112                   (sort dest #'string<))))
1113         (setq key (intern (format "%s@%s" 'ideographic-radical domain)))
1114         (when (and (memq key attributes)
1115                    (setq value (get-char-attribute char key)))
1116           (setq radical value)
1117           (insert (format "(%s . %S)\t; %c%s"
1118                           key
1119                           radical
1120                           (ideographic-radical radical)
1121                           line-breaking))
1122           (setq attributes (delq key attributes))
1123           )
1124         (setq key (intern (format "%s@%s" 'ideographic-strokes domain)))
1125         (when (and (memq key attributes)
1126                    (setq value (get-char-attribute char key)))
1127           (setq strokes value)
1128           (insert (format "(%s . %S)%s"
1129                           key
1130                           strokes
1131                           line-breaking))
1132           (setq attributes (delq key attributes))
1133           )
1134         (setq key (intern (format "%s@%s" 'total-strokes domain)))
1135         (when (and (memq key attributes)
1136                    (setq value (get-char-attribute char key)))
1137           (insert (format "(%s       . %S)%s"
1138                           key
1139                           value
1140                           line-breaking))
1141           (setq attributes (delq key attributes))
1142           )
1143         (dolist (feature '(ideographic-radical
1144                            ideographic-strokes
1145                            total-strokes))
1146           (setq key (intern (format "%s@%s*sources" feature domain)))
1147           (when (and (memq key attributes)
1148                      (setq value (get-char-attribute char key)))
1149             (insert (format "(%s%s" key line-breaking))
1150             (dolist (cell value)
1151               (insert (format " %s" cell)))
1152             (insert ")")
1153             (insert line-breaking)
1154             (setq attributes (delq key attributes))
1155             ))
1156         ))
1157     (when (and (memq 'ideographic-strokes attributes)
1158                (setq value (get-char-attribute char 'ideographic-strokes)))
1159       (setq strokes value)
1160       (insert (format "(ideographic-strokes . %S)%s"
1161                       strokes
1162                       line-breaking))
1163       (setq attributes (delq 'ideographic-strokes attributes))
1164       )
1165     (when (and (memq 'kangxi-radical attributes)
1166                (setq value (get-char-attribute char 'kangxi-radical)))
1167       (unless (eq value radical)
1168         (insert (format "(kangxi-radical\t . %S)\t; %c%s"
1169                         value
1170                         (ideographic-radical value)
1171                         line-breaking))
1172         (or radical
1173             (setq radical value)))
1174       (setq attributes (delq 'kangxi-radical attributes))
1175       )
1176     (when (and (memq 'kangxi-strokes attributes)
1177                (setq value (get-char-attribute char 'kangxi-strokes)))
1178       (unless (eq value strokes)
1179         (insert (format "(kangxi-strokes\t . %S)%s"
1180                         value
1181                         line-breaking))
1182         (or strokes
1183             (setq strokes value)))
1184       (setq attributes (delq 'kangxi-strokes attributes))
1185       )
1186     ;; (when (and (memq 'japanese-radical attributes)
1187     ;;            (setq value (get-char-attribute char 'japanese-radical)))
1188     ;;   (unless (eq value radical)
1189     ;;     (insert (format "(japanese-radical\t . %S)\t; %c%s"
1190     ;;                     value
1191     ;;                     (ideographic-radical value)
1192     ;;                     line-breaking))
1193     ;;     (or radical
1194     ;;         (setq radical value)))
1195     ;;   (setq attributes (delq 'japanese-radical attributes))
1196     ;;   )
1197     (when (and (memq 'japanese-strokes attributes)
1198                (setq value (get-char-attribute char 'japanese-strokes)))
1199       (unless (eq value strokes)
1200         (insert (format "(japanese-strokes\t . %S)%s"
1201                         value
1202                         line-breaking))
1203         (or strokes
1204             (setq strokes value)))
1205       (setq attributes (delq 'japanese-strokes attributes))
1206       )
1207     (when (and (memq 'cns-radical attributes)
1208                (setq value (get-char-attribute char 'cns-radical)))
1209       (insert (format "(cns-radical\t . %S)\t; %c%s"
1210                       value
1211                       (ideographic-radical value)
1212                       line-breaking))
1213       (setq attributes (delq 'cns-radical attributes))
1214       )
1215     (when (and (memq 'cns-strokes attributes)
1216                (setq value (get-char-attribute char 'cns-strokes)))
1217       (unless (eq value strokes)
1218         (insert (format "(cns-strokes\t . %S)%s"
1219                         value
1220                         line-breaking))
1221         (or strokes
1222             (setq strokes value)))
1223       (setq attributes (delq 'cns-strokes attributes))
1224       )
1225     ;; (when (and (memq 'shinjigen-1-radical attributes)
1226     ;;            (setq value (get-char-attribute char 'shinjigen-1-radical)))
1227     ;;   (unless (eq value radical)
1228     ;;     (insert (format "(shinjigen-1-radical . %S)\t; %c%s"
1229     ;;                     value
1230     ;;                     (ideographic-radical value)
1231     ;;                     line-breaking))
1232     ;;     (or radical
1233     ;;         (setq radical value)))
1234     ;;   (setq attributes (delq 'shinjigen-1-radical attributes))
1235     ;;   )
1236     ;; (when (and (memq 'ideographic- attributes)
1237     ;;            (setq value (get-char-attribute char 'ideographic-)))
1238     ;;   (insert "(ideographic-       ")
1239     ;;   (setq lbs (concat "\n" (make-string (current-column) ?\ ))
1240     ;;         separator nil)
1241     ;;   (while (consp value)
1242     ;;     (setq cell (car value))
1243     ;;     (if (integerp cell)
1244     ;;         (setq cell (decode-char '=ucs cell)))
1245     ;;     (cond ((characterp cell)
1246     ;;            (if separator
1247     ;;                (insert lbs))
1248     ;;            (if readable
1249     ;;                (insert (format "%S" cell))
1250     ;;              (char-db-insert-char-spec cell readable))
1251     ;;            (setq separator lbs))
1252     ;;           ((consp cell)
1253     ;;            (if separator
1254     ;;                (insert lbs))
1255     ;;            (if (consp (car cell))
1256     ;;                (char-db-insert-char-spec cell readable)
1257     ;;              (char-db-insert-char-reference cell readable))
1258     ;;            (setq separator lbs))
1259     ;;           (t
1260     ;;            (if separator
1261     ;;                (insert separator))
1262     ;;            (insert (prin1-to-string cell))
1263     ;;            (setq separator " ")))
1264     ;;     (setq value (cdr value)))
1265     ;;   (insert ")")
1266     ;;   (insert line-breaking)
1267     ;;   (setq attributes (delq 'ideographic- attributes)))
1268     (when (and (memq 'total-strokes attributes)
1269                (setq value (get-char-attribute char 'total-strokes)))
1270       (insert (format "(total-strokes       . %S)%s"
1271                       value
1272                       line-breaking))
1273       (setq attributes (delq 'total-strokes attributes))
1274       )
1275     ;; (when (and (memq '->ideograph attributes)
1276     ;;            (setq value (get-char-attribute char '->ideograph)))
1277     ;;   (insert (format "(->ideograph\t%s)%s"
1278     ;;                   (mapconcat (lambda (code)
1279     ;;                                (cond ((symbolp code)
1280     ;;                                       (symbol-name code))
1281     ;;                                      ((integerp code)
1282     ;;                                       (format "#x%04X" code))
1283     ;;                                      (t
1284     ;;                                       (format "%s %S"
1285     ;;                                               line-breaking code))))
1286     ;;                              value " ")
1287     ;;                   line-breaking))
1288     ;;   (setq attributes (delq '->ideograph attributes))
1289     ;;   )
1290     ;; (when (and (memq '->decomposition attributes)
1291     ;;            (setq value (get-char-attribute char '->decomposition)))
1292     ;;   (insert (format "(->decomposition\t%s)%s"
1293     ;;                   (mapconcat (lambda (code)
1294     ;;                                (cond ((symbolp code)
1295     ;;                                       (symbol-name code))
1296     ;;                                      ((characterp code)
1297     ;;                                       (if readable
1298     ;;                                           (format "%S" code)
1299     ;;                                         (format "#x%04X"
1300     ;;                                                 (char-int code))
1301     ;;                                         ))
1302     ;;                                      ((integerp code)
1303     ;;                                       (format "#x%04X" code))
1304     ;;                                      (t
1305     ;;                                       (format "%s%S" line-breaking code))))
1306     ;;                              value " ")
1307     ;;                   line-breaking))
1308     ;;   (setq attributes (delq '->decomposition attributes))
1309     ;;   )
1310     (if (equal (get-char-attribute char '->titlecase)
1311                (get-char-attribute char '->uppercase))
1312         (setq attributes (delq '->titlecase attributes)))
1313     (when (and (memq '->mojikyo attributes)
1314                (setq value (get-char-attribute char '->mojikyo)))
1315       (insert (format "(->mojikyo\t\t. %06d)\t; %c%s"
1316                       value (decode-char 'mojikyo value)
1317                       line-breaking))
1318       (setq attributes (delq '->mojikyo attributes))
1319       )
1320     ;; (when (and (memq 'hanyu-dazidian-vol attributes)
1321     ;;            (setq value (get-char-attribute char 'hanyu-dazidian-vol)))
1322     ;;   (insert (format "(hanyu-dazidian-vol  . %d)%s"
1323     ;;                   value line-breaking))
1324     ;;   (setq attributes (delq 'hanyu-dazidian-vol attributes))
1325     ;;   )
1326     ;; (when (and (memq 'hanyu-dazidian-page attributes)
1327     ;;            (setq value (get-char-attribute char 'hanyu-dazidian-page)))
1328     ;;   (insert (format "(hanyu-dazidian-page . %d)%s"
1329     ;;                   value line-breaking))
1330     ;;   (setq attributes (delq 'hanyu-dazidian-page attributes))
1331     ;;   )
1332     ;; (when (and (memq 'hanyu-dazidian-char attributes)
1333     ;;            (setq value (get-char-attribute char 'hanyu-dazidian-char)))
1334     ;;   (insert (format "(hanyu-dazidian-char . %d)%s"
1335     ;;                   value line-breaking))
1336     ;;   (setq attributes (delq 'hanyu-dazidian-char attributes))
1337     ;;   )
1338     (unless readable
1339       (dolist (ignored '(composition
1340                          ->denotational <-subsumptive ->ucs-unified
1341                          ->ideographic-component-forms))
1342         (setq attributes (delq ignored attributes))))
1343     (while attributes
1344       (setq name (car attributes))
1345       (unless (eq (setq value (get-char-attribute char name 'value-is-empty))
1346                   'value-is-empty)
1347         (cond ((setq ret (find-charset name))
1348                (setq name (charset-name ret))
1349                (when (not (memq name dest-ccss))
1350                  (setq dest-ccss (cons name dest-ccss))
1351                  (char-db-insert-ccs-feature name value line-breaking))
1352                )
1353               ((string-match "^=>ucs@" (symbol-name name))
1354                (insert (format "(%-18s . #x%04X)\t; %c%s"
1355                                name value (decode-char '=ucs value)
1356                                line-breaking))
1357                )
1358               ((eq name 'jisx0208-1978/4X)
1359                (insert (format "(%-18s . #x%04X)%s"
1360                                name value
1361                                line-breaking))
1362                )
1363               ((and
1364                 (not readable)
1365                 (not (eq name '->subsumptive))
1366                 (not (eq name '->uppercase))
1367                 (not (eq name '->lowercase))
1368                 (not (eq name '->titlecase))
1369                 (not (eq name '->canonical))
1370                 (not (eq name '->Bopomofo))
1371                 (not (eq name '->mistakable))
1372                 (not (eq name '->ideographic-variants))
1373                 (null (get-char-attribute
1374                        char (intern (format "%s*sources" name))))
1375                 (not (string-match "\\*sources$" (symbol-name name)))
1376                 (null (get-char-attribute
1377                        char (intern (format "%s*note" name))))
1378                 (not (string-match "\\*note$" (symbol-name name)))
1379                 (or (eq name '<-identical)
1380                     (eq name '<-uppercase)
1381                     (eq name '<-lowercase)
1382                     (eq name '<-titlecase)
1383                     (eq name '<-canonical)
1384                     (eq name '<-ideographic-variants)
1385                     ;; (eq name '<-synonyms)
1386                     (string-match "^<-synonyms" (symbol-name name))
1387                     (eq name '<-mistakable)
1388                     (when (string-match "^->" (symbol-name name))
1389                       (cond
1390                        ((string-match "^->fullwidth" (symbol-name name))
1391                         (not (and (consp value)
1392                                   (characterp (car value))
1393                                   (encode-char
1394                                    (car value) '=ucs 'defined-only)))
1395                         )
1396                        (t)))
1397                     ))
1398                )
1399               ((or (eq name 'ideographic-structure)
1400                    (eq name 'ideographic-combination)
1401                    ;; (eq name 'ideographic-)
1402                    (eq name '=decomposition)
1403                    (char-feature-base-name= '=decomposition name)
1404                    (char-feature-base-name= '=>decomposition name)
1405                    ;; (string-match "^=>*decomposition\\(@[^*]+\\)?$"
1406                    ;;               (symbol-name name))
1407                    (string-match "^\\(->\\|<-\\)[^*]*$" (symbol-name name))
1408                    (string-match "^\\(->\\|<-\\)[^*]*\\*sources$"
1409                                  (symbol-name name))
1410                    )
1411                (char-db-insert-relation-feature char name value
1412                                                 line-breaking
1413                                                 ccss readable))
1414               ;; ((memq name '(ideograph=
1415               ;;               original-ideograph-of
1416               ;;               ancient-ideograph-of
1417               ;;               vulgar-ideograph-of
1418               ;;               wrong-ideograph-of
1419               ;;               ;; simplified-ideograph-of
1420               ;;               ideographic-variants
1421               ;;               ;; ideographic-different-form-of
1422               ;;               ))
1423               ;;  (insert (format "(%-18s%s " name line-breaking))
1424               ;;  (setq lbs (concat "\n" (make-string (current-column) ?\ ))
1425               ;;        separator nil)
1426               ;;  (while (consp value)
1427               ;;    (setq cell (car value))
1428               ;;    (if (and (consp cell)
1429               ;;             (consp (car cell)))
1430               ;;        (progn
1431               ;;          (if separator
1432               ;;              (insert lbs))
1433               ;;          (char-db-insert-alist cell readable)
1434               ;;          (setq separator lbs))
1435               ;;      (if separator
1436               ;;          (insert separator))
1437               ;;      (insert (prin1-to-string cell))
1438               ;;      (setq separator " "))
1439               ;;    (setq value (cdr value)))
1440               ;;  (insert ")")
1441               ;;  (insert line-breaking))
1442               ((consp value)
1443                (insert (format "(%-18s " name))
1444                (setq lbs (concat "\n" (make-string (current-column) ?\ ))
1445                      separator nil)
1446                (while (consp value)
1447                  (setq cell (car value))
1448                  (if (and (consp cell)
1449                           (consp (car cell))
1450                           (setq ret (condition-case nil
1451                                         (find-char cell)
1452                                       (error nil))))
1453                      (progn
1454                        (setq rest cell
1455                              al nil
1456                              cal nil)
1457                        (while rest
1458                          (setq key (car (car rest)))
1459                          (if (find-charset key)
1460                              (setq cal (cons key cal))
1461                            (setq al (cons key al)))
1462                          (setq rest (cdr rest)))
1463                        (if separator
1464                            (insert lbs))
1465                        (insert-char-attributes ret
1466                                                readable
1467                                                al ; cal
1468                                                nil 'for-sub-node)
1469                        (setq separator lbs))
1470                    (setq ret (prin1-to-string cell))
1471                    (if separator
1472                        (if (< (+ (current-column)
1473                                  (length ret)
1474                                  (length separator))
1475                               76)
1476                            (insert separator)
1477                          (insert lbs)))
1478                    (insert ret)
1479                    (setq separator " "))
1480                  (setq value (cdr value)))
1481                (insert ")")
1482                (insert line-breaking))
1483               (t
1484                (insert (format "(%-18s" name))
1485                (setq ret (prin1-to-string value))
1486                (unless (< (+ (current-column)
1487                              (length ret)
1488                              3)
1489                           76)
1490                  (insert line-breaking))
1491                (insert " . " ret ")" line-breaking)
1492                ;; (insert (format "(%-18s . %S)%s"
1493                ;;                 name value
1494                ;;                 line-breaking))
1495                )
1496               ))
1497       (setq attributes (cdr attributes)))
1498     (insert ")")))
1499
1500 (defun insert-char-data (char &optional readable
1501                               attributes)
1502   (save-restriction
1503     (narrow-to-region (point)(point))
1504     (insert "(define-char
1505   '")
1506     (insert-char-attributes char readable attributes)
1507     (insert ")\n")
1508     (goto-char (point-min))
1509     (while (re-search-forward "[ \t]+$" nil t)
1510       (replace-match ""))
1511     ;; from tabify.
1512     (goto-char (point-min))
1513     (while (re-search-forward "[ \t][ \t][ \t]*" nil t)
1514       (let ((column (current-column))
1515             (indent-tabs-mode t))
1516         (delete-region (match-beginning 0) (point))
1517         (indent-to column)))
1518     (goto-char (point-max))
1519     ;; (tabify (point-min)(point-max))
1520     ))
1521
1522 (defun insert-char-data-with-variant (char &optional printable
1523                                            no-ucs-unified
1524                                            script excluded-script)
1525   (insert-char-data char printable)
1526   (let ((variants (char-variants char))
1527         rest
1528         variant vs ret)
1529     (setq variants (sort variants #'<))
1530     (setq rest variants)
1531     (setq variants (cons char variants))
1532     (while rest
1533       (setq variant (car rest))
1534       (unless (get-char-attribute variant '<-subsumptive)
1535         (if (and (or (null script)
1536                      (null (setq vs (get-char-attribute variant 'script)))
1537                      (memq script vs))
1538                  (or (null excluded-script)
1539                      (null (setq vs (get-char-attribute variant 'script)))
1540                      (not (memq excluded-script vs))))
1541             (unless (and no-ucs-unified (get-char-attribute variant '=ucs))
1542               (insert-char-data variant printable)
1543               (if (setq ret (char-variants variant))
1544                   (while ret
1545                     (or (memq (car ret) variants)
1546                         ;; (get-char-attribute (car ret) '<-subsumptive)
1547                         (setq rest (nconc rest (list (car ret)))))
1548                     (setq ret (cdr ret)))))))
1549       (setq rest (cdr rest)))))
1550
1551 (defun insert-char-range-data (min max &optional script excluded-script)
1552   (let ((code min)
1553         char)
1554     (while (<= code max)
1555       (setq char (decode-char '=ucs code))
1556       (if (encode-char char '=ucs 'defined-only)
1557           (insert-char-data-with-variant char nil 'no-ucs-unified
1558                                          script excluded-script))
1559       (setq code (1+ code)))))
1560
1561 (defun write-char-range-data-to-file (min max file
1562                                           &optional script excluded-script)
1563   (let ((coding-system-for-write char-db-file-coding-system))
1564     (with-temp-buffer
1565       (insert (format ";; -*- coding: %s -*-\n"
1566                       char-db-file-coding-system))
1567       (insert-char-range-data min max script excluded-script)
1568       (write-region (point-min)(point-max) file))))
1569
1570 (defvar what-character-original-window-configuration)
1571
1572 ;;;###autoload
1573 (defun what-char-definition (char)
1574   (interactive (list (char-after)))
1575   (let ((buf (get-buffer-create "*Character Description*"))
1576         (the-buf (current-buffer))
1577         (win-conf (current-window-configuration)))
1578     (pop-to-buffer buf)
1579     (make-local-variable 'what-character-original-window-configuration)
1580     (setq what-character-original-window-configuration win-conf)
1581     (setq buffer-read-only nil)
1582     (erase-buffer)
1583     (condition-case err
1584         (progn
1585           (insert-char-data-with-variant char 'printable)
1586           (unless (char-attribute-alist char)
1587             (insert (format ";; = %c\n"
1588                             (let* ((rest (split-char char))
1589                                    (ccs (pop rest))
1590                                    (code (pop rest)))
1591                               (while rest
1592                                 (setq code (logior (lsh code 8)
1593                                                    (pop rest))))
1594                               (decode-char ccs code)))))
1595           ;; (char-db-update-comment)
1596           (set-buffer-modified-p nil)
1597           (view-mode the-buf (lambda (buf)
1598                                (set-window-configuration
1599                                 what-character-original-window-configuration)
1600                                ))
1601           (goto-char (point-min)))
1602       (error (progn
1603                (set-window-configuration
1604                 what-character-original-window-configuration)
1605                (signal (car err) (cdr err)))))))
1606
1607
1608 ;;; @ end
1609 ;;;
1610
1611 (provide 'char-db-util)
1612
1613 ;;; char-db-util.el ends here