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     =>big5-cdp@component
 373     ==ucs@gb
 374     =ucs@gb
 375     ==ucs-var-002
 376     =ucs@JP/hanazono))
 377
 378
 379 ;;; @ char-db formatters
 380 ;;;
 381
 382 (defun char-db-make-char-spec (char)
 383   (let (ret char-spec)
 384     (cond ((characterp char)
 385            (cond ((and (setq ret (encode-char char '=ucs 'defined-only))
 386                        (not (and (<= #xE000 ret)(<= ret #xF8FF))))
 387                   (setq char-spec (list (cons '=ucs ret)))
 388                   (cond ((setq ret (get-char-attribute char 'name))
 389                          (setq char-spec (cons (cons 'name ret) char-spec))
 390                          )
 391                         ((setq ret (get-char-attribute char 'name*))
 392                          (setq char-spec (cons (cons 'name* ret) char-spec))
 393                          ))
 394                   )
 395                  ((encode-char char '=mj 'defined-only)
 396                   (setq char-spec nil)
 397                   (dolist (ccs (charset-list))
 398                     (if (and (or (eq ccs '=mj)
 399                                  ;; (eq (charset-property ccs 'iso-ir) 177)
 400                                  (string-match "=ucs@" (symbol-name ccs))
 401                                  )
 402                              (setq ccs (charset-name ccs))
 403                              (null (assq ccs char-spec))
 404                              (setq ret (encode-char char ccs 'defined-only)))
 405                         (setq char-spec (cons (cons ccs ret) char-spec))))
 406                   )
 407                  ((encode-char char '==mj 'defined-only)
 408                   (setq char-spec nil)
 409                   (dolist (ccs (charset-list))
 410                     (if (and (or (eq ccs '==mj)
 411                                  ;; (eq (charset-property ccs 'iso-ir) 177)
 412                                  (string-match "=ucs@" (symbol-name ccs))
 413                                  )
 414                              (setq ccs (charset-name ccs))
 415                              (null (assq ccs char-spec))
 416                              (setq ret (encode-char char ccs 'defined-only)))
 417                         (setq char-spec (cons (cons ccs ret) char-spec))))
 418                   )
 419                  ((encode-char char '=adobe-japan1 'defined-only)
 420                   (setq char-spec nil)
 421                   (dolist (ccs (charset-list))
 422                     (if (and (or (memq ccs
 423                                        '(=adobe-japan1-0
 424                                          =adobe-japan1-1
 425                                          =adobe-japan1-2
 426                                          =adobe-japan1-3
 427                                          =adobe-japan1-4
 428                                          =adobe-japan1-5
 429                                          =adobe-japan1-6
 430                                          ))
 431                                  ;; (eq (charset-property ccs 'iso-ir) 177)
 432                                  (string-match "=ucs@" (symbol-name ccs))
 433                                  )
 434                              (setq ccs (charset-name ccs))
 435                              (null (assq ccs char-spec))
 436                              (setq ret (encode-char char ccs 'defined-only)))
 437                         (setq char-spec (cons (cons ccs ret) char-spec))))
 438                   )
 439                  ((encode-char char '==adobe-japan1 'defined-only)
 440                   (setq char-spec nil)
 441                   (dolist (ccs (charset-list))
 442                     (if (and (or (memq ccs
 443                                        '(==adobe-japan1-0
 444                                          ==adobe-japan1-1
 445                                          ==adobe-japan1-2
 446                                          ==adobe-japan1-3
 447                                          ==adobe-japan1-4
 448                                          ==adobe-japan1-5
 449                                          ==adobe-japan1-6
 450                                          ))
 451                                  ;; (eq (charset-property ccs 'iso-ir) 177)
 452                                  (string-match "=ucs@" (symbol-name ccs))
 453                                  )
 454                              (setq ccs (charset-name ccs))
 455                              (null (assq ccs char-spec))
 456                              (setq ret (encode-char char ccs 'defined-only)))
 457                         (setq char-spec (cons (cons ccs ret) char-spec))))
 458                   )
 459                  ((setq ret
 460                         (catch 'tag
 461                           (let ((rest char-db-coded-charset-priority-list)
 462                                 ccs)
 463                             (while rest
 464                               (setq ccs (charset-name
 465                                          (find-charset (car rest))))
 466                               (if (setq ret
 467                                         (encode-char char ccs
 468                                                      'defined-only))
 469                                   (throw 'tag (cons ccs ret)))
 470                               (setq rest (cdr rest))))))
 471                   (setq char-spec (list ret))
 472                   (dolist (ccs (delq (car ret) (charset-list)))
 473                     (if (and (or (charset-iso-final-char ccs)
 474                                  (memq ccs
 475                                        '(=daikanwa
 476                                          =daikanwa@rev2
 477                                          ;; =gt-k
 478                                          =jis-x0208@1997
 479                                          ))
 480                                  (and (eq (charset-property ccs 'iso-ir) 177)
 481                                       (not (eq (charset-name ccs) '=ucs@big5)))
 482                                  ;; (string-match "=ucs@" (symbol-name ccs))
 483                                  )
 484                              (setq ccs (charset-name ccs))
 485                              (null (assq ccs char-spec))
 486                              (setq ret (encode-char char ccs 'defined-only)))
 487                         (setq char-spec (cons (cons ccs ret) char-spec))))
 488                   (if (null char-spec)
 489                       (setq char-spec (split-char char)))
 490                   (cond ((setq ret (get-char-attribute char 'name))
 491                          (setq char-spec (cons (cons 'name ret) char-spec))
 492                          )
 493                         ((setq ret (get-char-attribute char 'name*))
 494                          (setq char-spec (cons (cons 'name* ret) char-spec))
 495                          ))
 496                   )
 497                  ((setq ret (get-char-attribute
 498                              char 'ideographic-combination))
 499                   (setq char-spec
 500                         (cons (cons 'ideographic-combination ret)
 501                               char-spec))
 502                   ))
 503            char-spec)
 504           ((consp char)
 505            char))))
 506
 507 (defun char-db-insert-char-spec (char &optional readable column
 508                                       required-features)
 509   (unless column
 510     (setq column (current-column)))
 511   (let (char-spec temp-char)
 512     (setq char-spec (char-db-make-char-spec char))
 513     (unless (or (characterp char) ; char
 514                 (condition-case nil
 515                     (setq char (find-char char-spec))
 516                   (error nil)))
 517       ;; define temporary character
 518       ;;   Current implementation is dirty.
 519       (setq temp-char (define-char (cons '(ideograph-daikanwa . 0)
 520                                          char-spec)))
 521       (remove-char-attribute temp-char 'ideograph-daikanwa)
 522       (setq char temp-char))
 523     (insert-char-attributes char
 524                             readable
 525                             (union (mapcar #'car char-spec)
 526                                    required-features)
 527                             nil 'for-sub-node)
 528     (when temp-char
 529       ;; undefine temporary character
 530       ;;   Current implementation is dirty.
 531       (setq char-spec (char-attribute-alist temp-char))
 532       (while char-spec
 533         (remove-char-attribute temp-char (car (car char-spec)))
 534         (setq char-spec (cdr char-spec))))))
 535
 536 (defun char-db-insert-alist (alist &optional readable column)
 537   (unless column
 538     (setq column (current-column)))
 539   (let ((line-breaking
 540          (concat "\n" (make-string (1+ column) ?\ )))
 541         name value
 542         ret al ; cal
 543         key
 544         lbs cell rest separator)
 545     (insert "(")
 546     (while alist
 547       (setq name (car (car alist))
 548             value (cdr (car alist)))
 549       (cond ((eq name 'char)
 550              (insert "(char . ")
 551              (if (setq ret (condition-case nil
 552                                (find-char value)
 553                              (error nil)))
 554                  (progn
 555                    (setq al nil
 556                          ;; cal nil
 557                          )
 558                    (while value
 559                      (setq key (car (car value)))
 560                      ;; (if (find-charset key)
 561                      ;;     (setq cal (cons key cal))
 562                      (setq al (cons key al))
 563                      ;; )
 564                      (setq value (cdr value)))
 565                    (insert-char-attributes ret
 566                                            readable
 567                                            (or al 'none) ; cal
 568                                            nil 'for-sub-node))
 569                (insert (prin1-to-string value)))
 570              (insert ")")
 571              (insert line-breaking))
 572             ((consp value)
 573              (insert (format "(%-18s " name))
 574              (setq lbs (concat "\n" (make-string (current-column) ?\ )))
 575              (while (consp value)
 576                (setq cell (car value))
 577                (if (and (consp cell)
 578                         (consp (car cell))
 579                         (setq ret (condition-case nil
 580                                       (find-char cell)
 581                                     (error nil)))
 582                         )
 583                    (progn
 584                      (setq rest cell
 585                            al nil
 586                            ;; cal nil
 587                            )
 588                      (while rest
 589                        (setq key (car (car rest)))
 590                        ;; (if (find-charset key)
 591                        ;;     (setq cal (cons key cal))
 592                        (setq al (cons key al))
 593                        ;; )
 594                        (setq rest (cdr rest)))
 595                      (if separator
 596                          (insert lbs))
 597                      (insert-char-attributes ret
 598                                              readable
 599                                              al ; cal
 600                                              nil 'for-sub-node)
 601                      (setq separator lbs))
 602                  (if separator
 603                      (insert separator))
 604                  (insert (prin1-to-string cell))
 605                  (setq separator " "))
 606                (setq value (cdr value)))
 607              (insert ")")
 608              (insert line-breaking))
 609             (t
 610              (insert (format "(%-18s . %S)%s"
 611                              name value
 612                              line-breaking))))
 613       (setq alist (cdr alist))))
 614   (insert ")"))
 615
 616 (defun char-db-insert-char-reference (plist &optional readable column)
 617   (unless column
 618     (setq column (current-column)))
 619   (let ((line-breaking
 620          (concat "\n" (make-string (1+ column) ?\ )))
 621         (separator "")
 622         name value)
 623     (insert "(")
 624     (while plist
 625       (setq name (pop plist))
 626       (setq value (pop plist))
 627       (cond ((eq name :char)
 628              (insert separator)
 629              (insert ":char\t")
 630              (cond ((numberp value)
 631                     (setq value (decode-char '=ucs value)))
 632                    ;; ((consp value)
 633                    ;;  (setq value (or (find-char value)
 634                    ;;                  value)))
 635                    )
 636              (char-db-insert-char-spec value readable)
 637              (insert line-breaking)
 638              (setq separator ""))
 639             ((eq name :radical)
 640              (insert (format "%s%s\t%d ; %c%s"
 641                              separator
 642                              name value
 643                              (ideographic-radical value)
 644                              line-breaking))
 645              (setq separator ""))
 646             (t
 647              (insert (format "%s%s\t%S" separator name value))
 648              (setq separator line-breaking)))
 649       ))
 650   (insert ")"))
 651
 652 (defun char-db-decode-isolated-char (ccs code-point)
 653   (let (ret)
 654     (setq ret
 655           (cond ((eq ccs 'arabic-iso8859-6)
 656                  (decode-char ccs code-point))
 657                 ;; ((eq ccs '=gt)
 658                 ;;  (decode-builtin-char '==gt code-point))
 659                 ((and (memq ccs '(=gt-pj-1
 660                                   =gt-pj-2
 661                                   =gt-pj-3
 662                                   =gt-pj-4
 663                                   =gt-pj-5
 664                                   =gt-pj-6
 665                                   =gt-pj-7
 666                                   =gt-pj-8
 667                                   =gt-pj-9
 668                                   =gt-pj-10
 669                                   =gt-pj-11))
 670                       (setq ret (decode-char ccs code-point))
 671                       (setq ret (encode-char ret '=gt 'defined-only)))
 672                  (decode-builtin-char '=gt ret))
 673                 (t
 674                  (decode-builtin-char ccs code-point))))
 675     (cond ((null ret)
 676            (or (decode-char ccs code-point)
 677                (define-char (list (cons ccs code-point)))))
 678           ((and (<= 0 (char-int ret))
 679                 (<= (char-int ret) #x1F))
 680            (decode-char '=ucs (+ #x2400 (char-int ret))))
 681           ((= (char-int ret) #x7F)
 682            ?\u2421)
 683           (t ret))))
 684
 685 (defvar char-db-convert-obsolete-format t)
 686
 687 (defun char-db-insert-ccs-feature (name value line-breaking)
 688   (cond
 689    ((integerp value)
 690     (insert
 691      (format
 692       (cond
 693        ((memq name '(=>iwds-1
 694                      ==shinjigen
 695                      =shinjigen
 696                      =shinjigen@1ed ==shinjigen@1ed
 697                      =shinjigen@rev ==shinjigen@rev
 698                      =shinjigen/+p@rev ==shinjigen/+p@rev
 699                      ===daikanwa/ho ==daikanwa/ho
 700                      =daikanwa/ho =>>daikanwa/ho =>daikanwa/ho))
 701         "(%-18s .  %04d)\t; %c")
 702        ((eq name '=shinjigen@1ed/24pr)
 703         "(%-18s . %04d)\t; %c")
 704        ((or
 705          (memq name
 706                '(===daikanwa
 707                  ==daikanwa =daikanwa =>>daikanwa =>daikanwa
 708                  =daikanwa@rev1 =daikanwa@rev2
 709                  =daikanwa/+p ==daikanwa/+p ===daikanwa/+p
 710                  =>>daikanwa/+p
 711                  =daikanwa/+2p =>>daikanwa/+2p
 712                  =gt ==gt ===gt
 713                  =>>gt =+>gt =>gt
 714                  =gt-k ==gt-k ===gt-k
 715                  =>>gt-k =>gt-k
 716                  =adobe-japan1-0 ==adobe-japan1-0 ===adobe-japan1-0
 717                  =adobe-japan1-1 ==adobe-japan1-1 ===adobe-japan1-1
 718                  =adobe-japan1-2 ==adobe-japan1-2 ===adobe-japan1-2
 719                  =adobe-japan1-3 ==adobe-japan1-3 ===adobe-japan1-3
 720                  =adobe-japan1-4 ==adobe-japan1-4 ===adobe-japan1-4
 721                  =adobe-japan1-5 ==adobe-japan1-5 ===adobe-japan1-5
 722                  =adobe-japan1-6 ==adobe-japan1-6 ===adobe-japan1-6
 723                  =>>adobe-japan1-0 =+>adobe-japan1-0
 724                  =>>adobe-japan1-1 =+>adobe-japan1-1
 725                  =>>adobe-japan1-2 =+>adobe-japan1-2
 726                  =>>adobe-japan1-3 =+>adobe-japan1-3
 727                  =>>adobe-japan1-4 =+>adobe-japan1-4
 728                  =>>adobe-japan1-5 =+>adobe-japan1-5
 729                  =>>adobe-japan1-6 =+>adobe-japan1-6
 730                  =>cbeta =cbeta =>>cbeta ==cbeta ===cbeta
 731                  =zinbun-oracle =>zinbun-oracle
 732                  ===hng-jou ===hng-keg ===hng-dng ===hng-mam
 733                  ===hng-drt ===hng-kgk ===hng-myz ===hng-kda
 734                  ===hng-khi ===hng-khm ===hng-hok ===hng-kyd ===hng-sok
 735                  ===hng-yhk ===hng-kak ===hng-kar ===hng-kae
 736                  ===hng-sys ===hng-tsu ===hng-tzj
 737                  ===hng-hos ===hng-nak ===hng-jhk
 738                  ===hng-hod ===hng-gok ===hng-ink ===hng-nto
 739                  ===hng-nkm ===hng-k24 ===hng-nkk
 740                  ===hng-kcc ===hng-kcj ===hng-kbk ===hng-sik
 741                  ===hng-skk ===hng-kyu ===hng-ksk ===hng-wan
 742                  ===hng-okd ===hng-wad ===hng-kmi
 743                  ===hng-zkd ===hng-doh ===hng-jyu
 744                  ===hng-tzs ===hng-kss ===hng-kyo
 745                  ===hng-smk))
 746          ;; (string-match "^=adobe-" (symbol-name name))
 747          )
 748         "(%-18s . %05d)\t; %c")
 749        ((memq name '(=hanyo-denshi/ks
 750                      ==hanyo-denshi/ks ===hanyo-denshi/ks
 751                      =>>hanyo-denshi/ks
 752                      =koseki ==koseki
 753                      =mj ==mj ===mj =>>mj =>mj
 754                      =zihai mojikyo))
 755         "(%-18s . %06d)\t; %c")
 756        ((memq name '(=hanyo-denshi/tk ==hanyo-denshi/tk))
 757         "(%-18s . %08d)\t; %c")
 758        ((>= (charset-dimension name) 2)
 759         "(%-18s . #x%04X)\t; %c")
 760        (t
 761         "(%-18s . #x%02X)\t; %c"))
 762       name
 763       (if (= (charset-iso-graphic-plane name) 1)
 764           (logior value
 765                   (cond ((= (charset-dimension name) 1)
 766                          #x80)
 767                         ((= (charset-dimension name) 2)
 768                          #x8080)
 769                         ((= (charset-dimension name) 3)
 770                          #x808080)
 771                         (t 0)))
 772         value)
 773       (char-db-decode-isolated-char name value)))
 774     (if (and (= (charset-chars name) 94)
 775              (= (charset-dimension name) 2))
 776         (insert (format " [%02d-%02d]"
 777                         (- (lsh value -8) 32)
 778                         (- (logand value 255) 32))))
 779     )
 780    (t
 781     (insert (format "(%-18s . %s)" name value))
 782     ))
 783   (insert line-breaking))
 784
 785 (defun char-db-insert-relation-feature (char name value line-breaking
 786                                              ccss readable)
 787   (insert (format "(%-18s%s " name line-breaking))
 788   (let ((lbs (concat "\n" (make-string (current-column) ?\ )))
 789         separator cell sources required-features
 790         ret)
 791     (while (consp value)
 792       (setq cell (car value))
 793       (if (integerp cell)
 794           (setq cell (decode-char '=ucs cell)))
 795       (cond
 796        ((eq name '->subsumptive)
 797         (when (or (not (some (lambda (atr)
 798                                (get-char-attribute cell atr))
 799                              char-db-ignored-attributes))
 800                   (some (lambda (ccs)
 801                           (encode-char cell ccs 'defined-only))
 802                         ccss))
 803           (if separator
 804               (insert lbs))
 805           (let ((char-db-ignored-attributes
 806                  (cons '<-subsumptive
 807                        char-db-ignored-attributes)))
 808             (insert-char-attributes cell readable nil nil 'for-sub-node))
 809           (setq separator lbs))
 810         )
 811        ((characterp cell)
 812         (setq sources
 813               (get-char-attribute
 814                char (intern (format "%s*sources" name))))
 815         (setq required-features nil)
 816         (dolist (source sources)
 817           (cond
 818            ((memq source '(JP
 819                            JP/Jouyou
 820                            shinjigen shinjigen@1ed shinjigen@rev))
 821             (setq required-features
 822                   (union required-features
 823                          '(=jis-x0208
 824                            =jis-x0208@1990
 825                            =jis-x0213-1@2000
 826                            =jis-x0213-1@2004
 827                            =jis-x0213-2
 828                            =jis-x0212
 829                            =jis-x0208@1983
 830                            =jis-x0208@1978
 831                            =shinjigen
 832                            =shinjigen@1ed
 833                            =shinjigen@rev
 834                            =shinjigen/+p@rev))))
 835            ((eq source 'CN)
 836             (setq required-features
 837                   (union required-features
 838                          '(=gb2312
 839                            =gb12345
 840                            =iso-ir165)))))
 841           (cond
 842            ((find-charset (setq ret (intern (format "=%s" source))))
 843             (setq required-features
 844                   (cons ret required-features)))
 845            (t (setq required-features
 846                     (cons source required-features)))))
 847         (cond ((string-match "@JP" (symbol-name name))
 848                (setq required-features
 849                      (union required-features
 850                             '(=jis-x0208
 851                               =jis-x0208@1990
 852                               =jis-x0213-1-2000
 853                               =jis-x0213-2-2000
 854                               =jis-x0212
 855                               =jis-x0208@1983
 856                               =jis-x0208@1978))))
 857               ((string-match "@CN" (symbol-name name))
 858                (setq required-features
 859                      (union required-features
 860                             '(=gb2312
 861                               =gb12345
 862                               =iso-ir165)))))
 863         (if separator
 864             (insert lbs))
 865         (if readable
 866             (insert (format "%S" cell))
 867           (char-db-insert-char-spec cell readable
 868                                     nil
 869                                     required-features))
 870         (setq separator lbs))
 871        ((consp cell)
 872         (if separator
 873             (insert lbs))
 874         (if (consp (car cell))
 875             (char-db-insert-char-spec cell readable)
 876           (char-db-insert-char-reference cell readable))
 877         (setq separator lbs))
 878        (t
 879         (if separator
 880             (insert separator))
 881         (insert (prin1-to-string cell))
 882         (setq separator " ")))
 883       (setq value (cdr value)))
 884     (insert ")")
 885     (insert line-breaking)))
 886
 887 (defun insert-char-attributes (char &optional readable attributes column
 888                                     for-sub-node)
 889   (unless column
 890     (setq column (current-column)))
 891   (let (name value ; has-long-ccs-name
 892         rest
 893         radical strokes
 894         (line-breaking
 895          (concat "\n" (make-string (1+ column) ?\ )))
 896         lbs cell separator ret
 897         key al cal
 898         dest-ccss ; sources required-features
 899         ccss)
 900     (let (atr-d)
 901       (setq attributes
 902             (sort (if attributes
 903                       (if (consp attributes)
 904                           (progn
 905                             (dolist (name attributes)
 906                               (unless (memq name char-db-ignored-attributes)
 907                                 (if (find-charset name)
 908                                     (push name ccss))
 909                                 (push name atr-d)))
 910                             atr-d))
 911                     (dolist (name (char-attribute-list))
 912                       (unless (memq name char-db-ignored-attributes)
 913                         (if (find-charset name)
 914                             (push name ccss))
 915                         (push name atr-d)))
 916                     atr-d)
 917                   #'char-attribute-name<)))
 918     (insert "(")
 919     (when (memq '<-subsumptive attributes)
 920       (when (or readable (not for-sub-node))
 921         (when (setq value (get-char-attribute char '<-subsumptive))
 922           (char-db-insert-relation-feature char '<-subsumptive value
 923                                            line-breaking
 924                                            ccss readable)))
 925       (setq attributes (delq '<-subsumptive attributes)))
 926     (when (and (memq '<-denotational attributes)
 927                (setq value (get-char-attribute char '<-denotational)))
 928       (char-db-insert-relation-feature char '<-denotational value
 929                                        line-breaking
 930                                        ccss readable)
 931       (setq attributes (delq '<-denotational attributes)))
 932     (when (and (memq '<-denotational@component attributes)
 933                (setq value (get-char-attribute char '<-denotational@component)))
 934       (char-db-insert-relation-feature char '<-denotational@component value
 935                                        line-breaking
 936                                        ccss readable)
 937       (setq attributes (delq '<-denotational@component attributes)))
 938     (when (and (memq 'name attributes)
 939                (setq value (get-char-attribute char 'name)))
 940       (insert (format
 941                (if (> (+ (current-column) (length value)) 48)
 942                    "(name . %S)%s"
 943                  "(name               . %S)%s")
 944                value line-breaking))
 945       (setq attributes (delq 'name attributes))
 946       )
 947     (when (and (memq 'name* attributes)
 948                (setq value (get-char-attribute char 'name*)))
 949       (insert (format
 950                (if (> (+ (current-column) (length value)) 48)
 951                    "(name* . %S)%s"
 952                  "(name*              . %S)%s")
 953                value line-breaking))
 954       (setq attributes (delq 'name* attributes))
 955       )
 956     (when (and (memq 'script attributes)
 957                (setq value (get-char-attribute char 'script)))
 958       (insert (format "(script\t\t%s)%s"
 959                       (mapconcat (function prin1-to-string)
 960                                  value " ")
 961                       line-breaking))
 962       (setq attributes (delq 'script attributes))
 963       )
 964     (dolist (name '(=>ucs =>ucs*))
 965       (when (and (memq name attributes)
 966                  (setq value (get-char-attribute char name)))
 967         (insert (format "(%-18s . #x%04X)\t; %c%s"
 968                         name value (decode-char '=ucs value)
 969                         line-breaking))
 970         (setq attributes (delq name attributes))))
 971     (dolist (name '(=>ucs@gb =>ucs@big5))
 972       (when (and (memq name attributes)
 973                  (setq value (get-char-attribute char name)))
 974         (insert (format "(%-18s . #x%04X)\t; %c%s"
 975                         name value
 976                         (decode-char (intern
 977                                       (concat "="
 978                                               (substring
 979                                                (symbol-name name) 2)))
 980                                      value)
 981                         line-breaking))
 982         (setq attributes (delq name attributes))
 983         ))
 984     ;; (dolist (name '(=>daikanwa))
 985     ;;   (when (and (memq name attributes)
 986     ;;              (setq value (get-char-attribute char name)))
 987     ;;     (insert
 988     ;;      (if (integerp value)
 989     ;;          (format "(%-18s . %05d)\t; %c%s"
 990     ;;                  name value (decode-char '=daikanwa value)
 991     ;;                  line-breaking)
 992     ;;        (format "(%-18s %s)\t; %c%s"
 993     ;;                name
 994     ;;                (mapconcat (function prin1-to-string)
 995     ;;                           value " ")
 996     ;;                (char-representative-of-daikanwa char)
 997     ;;                line-breaking)))
 998     ;;     (setq attributes (delq name attributes))))
 999     (when (and (memq 'general-category attributes)
1000                (setq value (get-char-attribute char 'general-category)))
1001       (insert (format
1002                "(general-category\t%s) ; %s%s"
1003                (mapconcat (lambda (cell)
1004                             (format "%S" cell))
1005                           value " ")
1006                (cond ((rassoc value unidata-normative-category-alist)
1007                       "Normative Category")
1008                      ((rassoc value unidata-informative-category-alist)
1009                       "Informative Category")
1010                      (t
1011                       "Unknown Category"))
1012                line-breaking))
1013       (setq attributes (delq 'general-category attributes))
1014       )
1015     (when (and (memq 'bidi-category attributes)
1016                (setq value (get-char-attribute char 'bidi-category)))
1017       (insert (format "(bidi-category\t. %S)%s"
1018                       value
1019                       line-breaking))
1020       (setq attributes (delq 'bidi-category attributes))
1021       )
1022     (unless (or (not (memq 'mirrored attributes))
1023                 (eq (setq value (get-char-attribute char 'mirrored 'empty))
1024                     'empty))
1025       (insert (format "(mirrored\t\t. %S)%s"
1026                       value
1027                       line-breaking))
1028       (setq attributes (delq 'mirrored attributes))
1029       )
1030     (cond
1031      ((and (memq 'decimal-digit-value attributes)
1032            (setq value (get-char-attribute char 'decimal-digit-value)))
1033       (insert (format "(decimal-digit-value . %S)%s"
1034                       value
1035                       line-breaking))
1036       (setq attributes (delq 'decimal-digit-value attributes))
1037       (when (and (memq 'digit-value attributes)
1038                  (setq value (get-char-attribute char 'digit-value)))
1039         (insert (format "(digit-value\t . %S)%s"
1040                         value
1041                         line-breaking))
1042         (setq attributes (delq 'digit-value attributes))
1043         )
1044       (when (and (memq 'numeric-value attributes)
1045                  (setq value (get-char-attribute char 'numeric-value)))
1046         (insert (format "(numeric-value\t . %S)%s"
1047                         value
1048                         line-breaking))
1049         (setq attributes (delq 'numeric-value attributes))
1050         )
1051       )
1052      (t
1053       (when (and (memq 'digit-value attributes)
1054                  (setq value (get-char-attribute char 'digit-value)))
1055         (insert (format "(digit-value\t. %S)%s"
1056                         value
1057                         line-breaking))
1058         (setq attributes (delq 'digit-value attributes))
1059         )
1060       (when (and (memq 'numeric-value attributes)
1061                  (setq value (get-char-attribute char 'numeric-value)))
1062         (insert (format "(numeric-value\t. %S)%s"
1063                         value
1064                         line-breaking))
1065         (setq attributes (delq 'numeric-value attributes))
1066         )))
1067     (when (and (memq 'iso-10646-comment attributes)
1068                (setq value (get-char-attribute char 'iso-10646-comment)))
1069       (insert (format "(iso-10646-comment\t. %S)%s"
1070                       value
1071                       line-breaking))
1072       (setq attributes (delq 'iso-10646-comment attributes))
1073       )
1074     (when (and (memq 'morohashi-daikanwa attributes)
1075                (setq value (get-char-attribute char 'morohashi-daikanwa)))
1076       (insert (format "(morohashi-daikanwa\t%s)%s"
1077                       (mapconcat (function prin1-to-string) value " ")
1078                       line-breaking))
1079       (setq attributes (delq 'morohashi-daikanwa attributes))
1080       )
1081     (setq radical nil
1082           strokes nil)
1083     (when (and (memq 'ideographic-radical attributes)
1084                (setq value (get-char-attribute char 'ideographic-radical)))
1085       (setq radical value)
1086       (insert (format "(ideographic-radical . %S)\t; %c%s"
1087                       radical
1088                       (ideographic-radical radical)
1089                       line-breaking))
1090       (setq attributes (delq 'ideographic-radical attributes))
1091       )
1092     (when (and (memq 'shuowen-radical attributes)
1093                (setq value (get-char-attribute char 'shuowen-radical)))
1094       (insert (format "(shuowen-radical\t. %S)\t; %c%s"
1095                       value
1096                       (shuowen-radical value)
1097                       line-breaking))
1098       (setq attributes (delq 'shuowen-radical attributes))
1099       )
1100     (let (key)
1101       (dolist (domain
1102                (append
1103                 char-db-feature-domains
1104                 (let (dest domain)
1105                   (dolist (feature (char-attribute-list))
1106                     (setq feature (symbol-name feature))
1107                     (when (string-match
1108                            "\\(radical\\|strokes\\)@\\([^@*]+\\)\\(\\*\\|$\\)"
1109                            feature)
1110                       (setq domain (intern (match-string 2 feature)))
1111                      (unless (memq domain dest)
1112                        (setq dest (cons domain dest)))))
1113                   (sort dest #'string<))))
1114         (setq key (intern (format "%s@%s" 'ideographic-radical domain)))
1115         (when (and (memq key attributes)
1116                    (setq value (get-char-attribute char key)))
1117           (setq radical value)
1118           (insert (format "(%s . %S)\t; %c%s"
1119                           key
1120                           radical
1121                           (ideographic-radical radical)
1122                           line-breaking))
1123           (setq attributes (delq key attributes))
1124           )
1125         (setq key (intern (format "%s@%s" 'ideographic-strokes domain)))
1126         (when (and (memq key attributes)
1127                    (setq value (get-char-attribute char key)))
1128           (setq strokes value)
1129           (insert (format "(%s . %S)%s"
1130                           key
1131                           strokes
1132                           line-breaking))
1133           (setq attributes (delq key attributes))
1134           )
1135         (setq key (intern (format "%s@%s" 'total-strokes domain)))
1136         (when (and (memq key attributes)
1137                    (setq value (get-char-attribute char key)))
1138           (insert (format "(%s       . %S)%s"
1139                           key
1140                           value
1141                           line-breaking))
1142           (setq attributes (delq key attributes))
1143           )
1144         (dolist (feature '(ideographic-radical
1145                            ideographic-strokes
1146                            total-strokes))
1147           (setq key (intern (format "%s@%s*sources" feature domain)))
1148           (when (and (memq key attributes)
1149                      (setq value (get-char-attribute char key)))
1150             (insert (format "(%s%s" key line-breaking))
1151             (dolist (cell value)
1152               (insert (format " %s" cell)))
1153             (insert ")")
1154             (insert line-breaking)
1155             (setq attributes (delq key attributes))
1156             ))
1157         ))
1158     (when (and (memq 'ideographic-strokes attributes)
1159                (setq value (get-char-attribute char 'ideographic-strokes)))
1160       (setq strokes value)
1161       (insert (format "(ideographic-strokes . %S)%s"
1162                       strokes
1163                       line-breaking))
1164       (setq attributes (delq 'ideographic-strokes attributes))
1165       )
1166     (when (and (memq 'kangxi-radical attributes)
1167                (setq value (get-char-attribute char 'kangxi-radical)))
1168       (unless (eq value radical)
1169         (insert (format "(kangxi-radical\t . %S)\t; %c%s"
1170                         value
1171                         (ideographic-radical value)
1172                         line-breaking))
1173         (or radical
1174             (setq radical value)))
1175       (setq attributes (delq 'kangxi-radical attributes))
1176       )
1177     (when (and (memq 'kangxi-strokes attributes)
1178                (setq value (get-char-attribute char 'kangxi-strokes)))
1179       (unless (eq value strokes)
1180         (insert (format "(kangxi-strokes\t . %S)%s"
1181                         value
1182                         line-breaking))
1183         (or strokes
1184             (setq strokes value)))
1185       (setq attributes (delq 'kangxi-strokes attributes))
1186       )
1187     ;; (when (and (memq 'japanese-radical attributes)
1188     ;;            (setq value (get-char-attribute char 'japanese-radical)))
1189     ;;   (unless (eq value radical)
1190     ;;     (insert (format "(japanese-radical\t . %S)\t; %c%s"
1191     ;;                     value
1192     ;;                     (ideographic-radical value)
1193     ;;                     line-breaking))
1194     ;;     (or radical
1195     ;;         (setq radical value)))
1196     ;;   (setq attributes (delq 'japanese-radical attributes))
1197     ;;   )
1198     (when (and (memq 'japanese-strokes attributes)
1199                (setq value (get-char-attribute char 'japanese-strokes)))
1200       (unless (eq value strokes)
1201         (insert (format "(japanese-strokes\t . %S)%s"
1202                         value
1203                         line-breaking))
1204         (or strokes
1205             (setq strokes value)))
1206       (setq attributes (delq 'japanese-strokes attributes))
1207       )
1208     (when (and (memq 'cns-radical attributes)
1209                (setq value (get-char-attribute char 'cns-radical)))
1210       (insert (format "(cns-radical\t . %S)\t; %c%s"
1211                       value
1212                       (ideographic-radical value)
1213                       line-breaking))
1214       (setq attributes (delq 'cns-radical attributes))
1215       )
1216     (when (and (memq 'cns-strokes attributes)
1217                (setq value (get-char-attribute char 'cns-strokes)))
1218       (unless (eq value strokes)
1219         (insert (format "(cns-strokes\t . %S)%s"
1220                         value
1221                         line-breaking))
1222         (or strokes
1223             (setq strokes value)))
1224       (setq attributes (delq 'cns-strokes attributes))
1225       )
1226     ;; (when (and (memq 'shinjigen-1-radical attributes)
1227     ;;            (setq value (get-char-attribute char 'shinjigen-1-radical)))
1228     ;;   (unless (eq value radical)
1229     ;;     (insert (format "(shinjigen-1-radical . %S)\t; %c%s"
1230     ;;                     value
1231     ;;                     (ideographic-radical value)
1232     ;;                     line-breaking))
1233     ;;     (or radical
1234     ;;         (setq radical value)))
1235     ;;   (setq attributes (delq 'shinjigen-1-radical attributes))
1236     ;;   )
1237     ;; (when (and (memq 'ideographic- attributes)
1238     ;;            (setq value (get-char-attribute char 'ideographic-)))
1239     ;;   (insert "(ideographic-       ")
1240     ;;   (setq lbs (concat "\n" (make-string (current-column) ?\ ))
1241     ;;         separator nil)
1242     ;;   (while (consp value)
1243     ;;     (setq cell (car value))
1244     ;;     (if (integerp cell)
1245     ;;         (setq cell (decode-char '=ucs cell)))
1246     ;;     (cond ((characterp cell)
1247     ;;            (if separator
1248     ;;                (insert lbs))
1249     ;;            (if readable
1250     ;;                (insert (format "%S" cell))
1251     ;;              (char-db-insert-char-spec cell readable))
1252     ;;            (setq separator lbs))
1253     ;;           ((consp cell)
1254     ;;            (if separator
1255     ;;                (insert lbs))
1256     ;;            (if (consp (car cell))
1257     ;;                (char-db-insert-char-spec cell readable)
1258     ;;              (char-db-insert-char-reference cell readable))
1259     ;;            (setq separator lbs))
1260     ;;           (t
1261     ;;            (if separator
1262     ;;                (insert separator))
1263     ;;            (insert (prin1-to-string cell))
1264     ;;            (setq separator " ")))
1265     ;;     (setq value (cdr value)))
1266     ;;   (insert ")")
1267     ;;   (insert line-breaking)
1268     ;;   (setq attributes (delq 'ideographic- attributes)))
1269     (when (and (memq 'total-strokes attributes)
1270                (setq value (get-char-attribute char 'total-strokes)))
1271       (insert (format "(total-strokes       . %S)%s"
1272                       value
1273                       line-breaking))
1274       (setq attributes (delq 'total-strokes attributes))
1275       )
1276     ;; (when (and (memq '->ideograph attributes)
1277     ;;            (setq value (get-char-attribute char '->ideograph)))
1278     ;;   (insert (format "(->ideograph\t%s)%s"
1279     ;;                   (mapconcat (lambda (code)
1280     ;;                                (cond ((symbolp code)
1281     ;;                                       (symbol-name code))
1282     ;;                                      ((integerp code)
1283     ;;                                       (format "#x%04X" code))
1284     ;;                                      (t
1285     ;;                                       (format "%s %S"
1286     ;;                                               line-breaking code))))
1287     ;;                              value " ")
1288     ;;                   line-breaking))
1289     ;;   (setq attributes (delq '->ideograph attributes))
1290     ;;   )
1291     ;; (when (and (memq '->decomposition attributes)
1292     ;;            (setq value (get-char-attribute char '->decomposition)))
1293     ;;   (insert (format "(->decomposition\t%s)%s"
1294     ;;                   (mapconcat (lambda (code)
1295     ;;                                (cond ((symbolp code)
1296     ;;                                       (symbol-name code))
1297     ;;                                      ((characterp code)
1298     ;;                                       (if readable
1299     ;;                                           (format "%S" code)
1300     ;;                                         (format "#x%04X"
1301     ;;                                                 (char-int code))
1302     ;;                                         ))
1303     ;;                                      ((integerp code)
1304     ;;                                       (format "#x%04X" code))
1305     ;;                                      (t
1306     ;;                                       (format "%s%S" line-breaking code))))
1307     ;;                              value " ")
1308     ;;                   line-breaking))
1309     ;;   (setq attributes (delq '->decomposition attributes))
1310     ;;   )
1311     (if (equal (get-char-attribute char '->titlecase)
1312                (get-char-attribute char '->uppercase))
1313         (setq attributes (delq '->titlecase attributes)))
1314     (when (and (memq '->mojikyo attributes)
1315                (setq value (get-char-attribute char '->mojikyo)))
1316       (insert (format "(->mojikyo\t\t. %06d)\t; %c%s"
1317                       value (decode-char 'mojikyo value)
1318                       line-breaking))
1319       (setq attributes (delq '->mojikyo attributes))
1320       )
1321     ;; (when (and (memq 'hanyu-dazidian-vol attributes)
1322     ;;            (setq value (get-char-attribute char 'hanyu-dazidian-vol)))
1323     ;;   (insert (format "(hanyu-dazidian-vol  . %d)%s"
1324     ;;                   value line-breaking))
1325     ;;   (setq attributes (delq 'hanyu-dazidian-vol attributes))
1326     ;;   )
1327     ;; (when (and (memq 'hanyu-dazidian-page attributes)
1328     ;;            (setq value (get-char-attribute char 'hanyu-dazidian-page)))
1329     ;;   (insert (format "(hanyu-dazidian-page . %d)%s"
1330     ;;                   value line-breaking))
1331     ;;   (setq attributes (delq 'hanyu-dazidian-page attributes))
1332     ;;   )
1333     ;; (when (and (memq 'hanyu-dazidian-char attributes)
1334     ;;            (setq value (get-char-attribute char 'hanyu-dazidian-char)))
1335     ;;   (insert (format "(hanyu-dazidian-char . %d)%s"
1336     ;;                   value line-breaking))
1337     ;;   (setq attributes (delq 'hanyu-dazidian-char attributes))
1338     ;;   )
1339     (unless readable
1340       (dolist (ignored '(composition
1341                          ->denotational <-subsumptive ->ucs-unified
1342                          ->ideographic-component-forms))
1343         (setq attributes (delq ignored attributes))))
1344     (while attributes
1345       (setq name (car attributes))
1346       (unless (eq (setq value (get-char-attribute char name 'value-is-empty))
1347                   'value-is-empty)
1348         (cond ((setq ret (find-charset name))
1349                (setq name (charset-name ret))
1350                (when (not (memq name dest-ccss))
1351                  (setq dest-ccss (cons name dest-ccss))
1352                  (char-db-insert-ccs-feature name value line-breaking))
1353                )
1354               ((string-match "^=>ucs@" (symbol-name name))
1355                (insert (format "(%-18s . #x%04X)\t; %c%s"
1356                                name value (decode-char '=ucs value)
1357                                line-breaking))
1358                )
1359               ((eq name 'jisx0208-1978/4X)
1360                (insert (format "(%-18s . #x%04X)%s"
1361                                name value
1362                                line-breaking))
1363                )
1364               ((and
1365                 (not readable)
1366                 (not (eq name '->subsumptive))
1367                 (not (eq name '->uppercase))
1368                 (not (eq name '->lowercase))
1369                 (not (eq name '->titlecase))
1370                 (not (eq name '->canonical))
1371                 (not (eq name '->Bopomofo))
1372                 (not (eq name '->mistakable))
1373                 (not (eq name '->ideographic-variants))
1374                 (null (get-char-attribute
1375                        char (intern (format "%s*sources" name))))
1376                 (not (string-match "\\*sources$" (symbol-name name)))
1377                 (null (get-char-attribute
1378                        char (intern (format "%s*note" name))))
1379                 (not (string-match "\\*note$" (symbol-name name)))
1380                 (or (eq name '<-identical)
1381                     (eq name '<-uppercase)
1382                     (eq name '<-lowercase)
1383                     (eq name '<-titlecase)
1384                     (eq name '<-canonical)
1385                     (eq name '<-ideographic-variants)
1386                     ;; (eq name '<-synonyms)
1387                     (string-match "^<-synonyms" (symbol-name name))
1388                     (eq name '<-mistakable)
1389                     (when (string-match "^->" (symbol-name name))
1390                       (cond
1391                        ((string-match "^->fullwidth" (symbol-name name))
1392                         (not (and (consp value)
1393                                   (characterp (car value))
1394                                   (encode-char
1395                                    (car value) '=ucs 'defined-only)))
1396                         )
1397                        (t)))
1398                     ))
1399                )
1400               ((or (eq name 'ideographic-structure)
1401                    (eq name 'ideographic-combination)
1402                    ;; (eq name 'ideographic-)
1403                    (eq name '=decomposition)
1404                    (char-feature-base-name= '=decomposition name)
1405                    (char-feature-base-name= '=>decomposition name)
1406                    ;; (string-match "^=>*decomposition\\(@[^*]+\\)?$"
1407                    ;;               (symbol-name name))
1408                    (string-match "^\\(->\\|<-\\)[^*]*$" (symbol-name name))
1409                    (string-match "^\\(->\\|<-\\)[^*]*\\*sources$"
1410                                  (symbol-name name))
1411                    )
1412                (char-db-insert-relation-feature char name value
1413                                                 line-breaking
1414                                                 ccss readable))
1415               ;; ((memq name '(ideograph=
1416               ;;               original-ideograph-of
1417               ;;               ancient-ideograph-of
1418               ;;               vulgar-ideograph-of
1419               ;;               wrong-ideograph-of
1420               ;;               ;; simplified-ideograph-of
1421               ;;               ideographic-variants
1422               ;;               ;; ideographic-different-form-of
1423               ;;               ))
1424               ;;  (insert (format "(%-18s%s " name line-breaking))
1425               ;;  (setq lbs (concat "\n" (make-string (current-column) ?\ ))
1426               ;;        separator nil)
1427               ;;  (while (consp value)
1428               ;;    (setq cell (car value))
1429               ;;    (if (and (consp cell)
1430               ;;             (consp (car cell)))
1431               ;;        (progn
1432               ;;          (if separator
1433               ;;              (insert lbs))
1434               ;;          (char-db-insert-alist cell readable)
1435               ;;          (setq separator lbs))
1436               ;;      (if separator
1437               ;;          (insert separator))
1438               ;;      (insert (prin1-to-string cell))
1439               ;;      (setq separator " "))
1440               ;;    (setq value (cdr value)))
1441               ;;  (insert ")")
1442               ;;  (insert line-breaking))
1443               ((consp value)
1444                (insert (format "(%-18s " name))
1445                (setq lbs (concat "\n" (make-string (current-column) ?\ ))
1446                      separator nil)
1447                (while (consp value)
1448                  (setq cell (car value))
1449                  (if (and (consp cell)
1450                           (consp (car cell))
1451                           (setq ret (condition-case nil
1452                                         (find-char cell)
1453                                       (error nil))))
1454                      (progn
1455                        (setq rest cell
1456                              al nil
1457                              cal nil)
1458                        (while rest
1459                          (setq key (car (car rest)))
1460                          (if (find-charset key)
1461                              (setq cal (cons key cal))
1462                            (setq al (cons key al)))
1463                          (setq rest (cdr rest)))
1464                        (if separator
1465                            (insert lbs))
1466                        (insert-char-attributes ret
1467                                                readable
1468                                                al ; cal
1469                                                nil 'for-sub-node)
1470                        (setq separator lbs))
1471                    (setq ret (prin1-to-string cell))
1472                    (if separator
1473                        (if (< (+ (current-column)
1474                                  (length ret)
1475                                  (length separator))
1476                               76)
1477                            (insert separator)
1478                          (insert lbs)))
1479                    (insert ret)
1480                    (setq separator " "))
1481                  (setq value (cdr value)))
1482                (insert ")")
1483                (insert line-breaking))
1484               (t
1485                (insert (format "(%-18s" name))
1486                (setq ret (prin1-to-string value))
1487                (unless (< (+ (current-column)
1488                              (length ret)
1489                              3)
1490                           76)
1491                  (insert line-breaking))
1492                (insert " . " ret ")" line-breaking)
1493                ;; (insert (format "(%-18s . %S)%s"
1494                ;;                 name value
1495                ;;                 line-breaking))
1496                )
1497               ))
1498       (setq attributes (cdr attributes)))
1499     (insert ")")))
1500
1501 (defun insert-char-data (char &optional readable
1502                               attributes)
1503   (save-restriction
1504     (narrow-to-region (point)(point))
1505     (insert "(define-char
1506   '")
1507     (insert-char-attributes char readable attributes)
1508     (insert ")\n")
1509     (goto-char (point-min))
1510     (while (re-search-forward "[ \t]+$" nil t)
1511       (replace-match ""))
1512     ;; from tabify.
1513     (goto-char (point-min))
1514     (while (re-search-forward "[ \t][ \t][ \t]*" nil t)
1515       (let ((column (current-column))
1516             (indent-tabs-mode t))
1517         (delete-region (match-beginning 0) (point))
1518         (indent-to column)))
1519     (goto-char (point-max))
1520     ;; (tabify (point-min)(point-max))
1521     ))
1522
1523 (defun insert-char-data-with-variant (char &optional printable
1524                                            no-ucs-unified
1525                                            script excluded-script)
1526   (insert-char-data char printable)
1527   (let ((variants (char-variants char))
1528         rest
1529         variant vs ret)
1530     (setq variants (sort variants #'<))
1531     (setq rest variants)
1532     (setq variants (cons char variants))
1533     (while rest
1534       (setq variant (car rest))
1535       (unless (get-char-attribute variant '<-subsumptive)
1536         (if (and (or (null script)
1537                      (null (setq vs (get-char-attribute variant 'script)))
1538                      (memq script vs))
1539                  (or (null excluded-script)
1540                      (null (setq vs (get-char-attribute variant 'script)))
1541                      (not (memq excluded-script vs))))
1542             (unless (and no-ucs-unified (get-char-attribute variant '=ucs))
1543               (insert-char-data variant printable)
1544               (if (setq ret (char-variants variant))
1545                   (while ret
1546                     (or (memq (car ret) variants)
1547                         ;; (get-char-attribute (car ret) '<-subsumptive)
1548                         (setq rest (nconc rest (list (car ret)))))
1549                     (setq ret (cdr ret)))))))
1550       (setq rest (cdr rest)))))
1551
1552 (defun insert-char-range-data (min max &optional script excluded-script)
1553   (let ((code min)
1554         char)
1555     (while (<= code max)
1556       (setq char (decode-char '=ucs code))
1557       (if (encode-char char '=ucs 'defined-only)
1558           (insert-char-data-with-variant char nil 'no-ucs-unified
1559                                          script excluded-script))
1560       (setq code (1+ code)))))
1561
1562 (defun write-char-range-data-to-file (min max file
1563                                           &optional script excluded-script)
1564   (let ((coding-system-for-write char-db-file-coding-system))
1565     (with-temp-buffer
1566       (insert (format ";; -*- coding: %s -*-\n"
1567                       char-db-file-coding-system))
1568       (insert-char-range-data min max script excluded-script)
1569       (write-region (point-min)(point-max) file))))
1570
1571 (defvar what-character-original-window-configuration)
1572
1573 ;;;###autoload
1574 (defun what-char-definition (char)
1575   (interactive (list (char-after)))
1576   (let ((buf (get-buffer-create "*Character Description*"))
1577         (the-buf (current-buffer))
1578         (win-conf (current-window-configuration)))
1579     (pop-to-buffer buf)
1580     (make-local-variable 'what-character-original-window-configuration)
1581     (setq what-character-original-window-configuration win-conf)
1582     (setq buffer-read-only nil)
1583     (erase-buffer)
1584     (condition-case err
1585         (progn
1586           (insert-char-data-with-variant char 'printable)
1587           (unless (char-attribute-alist char)
1588             (insert (format ";; = %c\n"
1589                             (let* ((rest (split-char char))
1590                                    (ccs (pop rest))
1591                                    (code (pop rest)))
1592                               (while rest
1593                                 (setq code (logior (lsh code 8)
1594                                                    (pop rest))))
1595                               (decode-char ccs code)))))
1596           ;; (char-db-update-comment)
1597           (set-buffer-modified-p nil)
1598           (view-mode the-buf (lambda (buf)
1599                                (set-window-configuration
1600                                 what-character-original-window-configuration)
1601                                ))
1602           (goto-char (point-min)))
1603       (error (progn
1604                (set-window-configuration
1605                 what-character-original-window-configuration)
1606                (signal (car err) (cdr err)))))))
1607
1608
1609 ;;; @ end
1610 ;;;
1611
1612 (provide 'char-db-util)
1613
1614 ;;; char-db-util.el ends here