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     =big5-cdp-var-001
 359     =big5-cdp-var-002
 360     =big5-cdp-var-003
 361     =big5-cdp-var-004
 362     =big5-cdp-var-005
 363     =big5-cdp-itaiji-001
 364     =big5-cdp-itaiji-002
 365     =big5-cdp-itaiji-003
 366     =>big5-cdp-itaiji-001
 367     =>ucs@iwds-1
 368     =>ucs@cognate
 369     =>ucs@component
 370     ==ucs@gb
 371     =ucs@gb
 372     ==ucs-var-002
 373     =ucs@JP/hanazono))
 374
 375
 376 ;;; @ char-db formatters
 377 ;;;
 378
 379 (defun char-db-make-char-spec (char)
 380   (let (ret char-spec)
 381     (cond ((characterp char)
 382            (cond ((and (setq ret (encode-char char '=ucs 'defined-only))
 383                        (not (and (<= #xE000 ret)(<= ret #xF8FF))))
 384                   (setq char-spec (list (cons '=ucs ret)))
 385                   (cond ((setq ret (get-char-attribute char 'name))
 386                          (setq char-spec (cons (cons 'name ret) char-spec))
 387                          )
 388                         ((setq ret (get-char-attribute char 'name*))
 389                          (setq char-spec (cons (cons 'name* ret) char-spec))
 390                          ))
 391                   )
 392                  ((encode-char char '=mj 'defined-only)
 393                   (setq char-spec nil)
 394                   (dolist (ccs (charset-list))
 395                     (if (and (or (eq ccs '=mj)
 396                                  ;; (eq (charset-property ccs 'iso-ir) 177)
 397                                  (string-match "=ucs@" (symbol-name ccs))
 398                                  )
 399                              (setq ccs (charset-name ccs))
 400                              (null (assq ccs char-spec))
 401                              (setq ret (encode-char char ccs 'defined-only)))
 402                         (setq char-spec (cons (cons ccs ret) char-spec))))
 403                   )
 404                  ((encode-char char '==mj 'defined-only)
 405                   (setq char-spec nil)
 406                   (dolist (ccs (charset-list))
 407                     (if (and (or (eq ccs '==mj)
 408                                  ;; (eq (charset-property ccs 'iso-ir) 177)
 409                                  (string-match "=ucs@" (symbol-name ccs))
 410                                  )
 411                              (setq ccs (charset-name ccs))
 412                              (null (assq ccs char-spec))
 413                              (setq ret (encode-char char ccs 'defined-only)))
 414                         (setq char-spec (cons (cons ccs ret) char-spec))))
 415                   )
 416                  ((encode-char char '=adobe-japan1 'defined-only)
 417                   (setq char-spec nil)
 418                   (dolist (ccs (charset-list))
 419                     (if (and (or (memq ccs
 420                                        '(=adobe-japan1-0
 421                                          =adobe-japan1-1
 422                                          =adobe-japan1-2
 423                                          =adobe-japan1-3
 424                                          =adobe-japan1-4
 425                                          =adobe-japan1-5
 426                                          =adobe-japan1-6
 427                                          ))
 428                                  ;; (eq (charset-property ccs 'iso-ir) 177)
 429                                  (string-match "=ucs@" (symbol-name ccs))
 430                                  )
 431                              (setq ccs (charset-name ccs))
 432                              (null (assq ccs char-spec))
 433                              (setq ret (encode-char char ccs 'defined-only)))
 434                         (setq char-spec (cons (cons ccs ret) char-spec))))
 435                   )
 436                  ((encode-char char '==adobe-japan1 'defined-only)
 437                   (setq char-spec nil)
 438                   (dolist (ccs (charset-list))
 439                     (if (and (or (memq ccs
 440                                        '(==adobe-japan1-0
 441                                          ==adobe-japan1-1
 442                                          ==adobe-japan1-2
 443                                          ==adobe-japan1-3
 444                                          ==adobe-japan1-4
 445                                          ==adobe-japan1-5
 446                                          ==adobe-japan1-6
 447                                          ))
 448                                  ;; (eq (charset-property ccs 'iso-ir) 177)
 449                                  (string-match "=ucs@" (symbol-name ccs))
 450                                  )
 451                              (setq ccs (charset-name ccs))
 452                              (null (assq ccs char-spec))
 453                              (setq ret (encode-char char ccs 'defined-only)))
 454                         (setq char-spec (cons (cons ccs ret) char-spec))))
 455                   )
 456                  ((setq ret
 457                         (catch 'tag
 458                           (let ((rest char-db-coded-charset-priority-list)
 459                                 ccs)
 460                             (while rest
 461                               (setq ccs (charset-name
 462                                          (find-charset (car rest))))
 463                               (if (setq ret
 464                                         (encode-char char ccs
 465                                                      'defined-only))
 466                                   (throw 'tag (cons ccs ret)))
 467                               (setq rest (cdr rest))))))
 468                   (setq char-spec (list ret))
 469                   (dolist (ccs (delq (car ret) (charset-list)))
 470                     (if (and (or (charset-iso-final-char ccs)
 471                                  (memq ccs
 472                                        '(=daikanwa
 473                                          =daikanwa@rev2
 474                                          ;; =gt-k
 475                                          =jis-x0208@1997
 476                                          ))
 477                                  (and (eq (charset-property ccs 'iso-ir) 177)
 478                                       (not (eq (charset-name ccs) '=ucs@big5)))
 479                                  ;; (string-match "=ucs@" (symbol-name ccs))
 480                                  )
 481                              (setq ccs (charset-name ccs))
 482                              (null (assq ccs char-spec))
 483                              (setq ret (encode-char char ccs 'defined-only)))
 484                         (setq char-spec (cons (cons ccs ret) char-spec))))
 485                   (if (null char-spec)
 486                       (setq char-spec (split-char char)))
 487                   (cond ((setq ret (get-char-attribute char 'name))
 488                          (setq char-spec (cons (cons 'name ret) char-spec))
 489                          )
 490                         ((setq ret (get-char-attribute char 'name*))
 491                          (setq char-spec (cons (cons 'name* ret) char-spec))
 492                          ))
 493                   )
 494                  ((setq ret (get-char-attribute
 495                              char 'ideographic-combination))
 496                   (setq char-spec
 497                         (cons (cons 'ideographic-combination ret)
 498                               char-spec))
 499                   ))
 500            char-spec)
 501           ((consp char)
 502            char))))
 503
 504 (defun char-db-insert-char-spec (char &optional readable column
 505                                       required-features)
 506   (unless column
 507     (setq column (current-column)))
 508   (let (char-spec temp-char)
 509     (setq char-spec (char-db-make-char-spec char))
 510     (unless (or (characterp char) ; char
 511                 (condition-case nil
 512                     (setq char (find-char char-spec))
 513                   (error nil)))
 514       ;; define temporary character
 515       ;;   Current implementation is dirty.
 516       (setq temp-char (define-char (cons '(ideograph-daikanwa . 0)
 517                                          char-spec)))
 518       (remove-char-attribute temp-char 'ideograph-daikanwa)
 519       (setq char temp-char))
 520     (insert-char-attributes char
 521                             readable
 522                             (union (mapcar #'car char-spec)
 523                                    required-features)
 524                             nil 'for-sub-node)
 525     (when temp-char
 526       ;; undefine temporary character
 527       ;;   Current implementation is dirty.
 528       (setq char-spec (char-attribute-alist temp-char))
 529       (while char-spec
 530         (remove-char-attribute temp-char (car (car char-spec)))
 531         (setq char-spec (cdr char-spec))))))
 532
 533 (defun char-db-insert-alist (alist &optional readable column)
 534   (unless column
 535     (setq column (current-column)))
 536   (let ((line-breaking
 537          (concat "\n" (make-string (1+ column) ?\ )))
 538         name value
 539         ret al ; cal
 540         key
 541         lbs cell rest separator)
 542     (insert "(")
 543     (while alist
 544       (setq name (car (car alist))
 545             value (cdr (car alist)))
 546       (cond ((eq name 'char)
 547              (insert "(char . ")
 548              (if (setq ret (condition-case nil
 549                                (find-char value)
 550                              (error nil)))
 551                  (progn
 552                    (setq al nil
 553                          ;; cal nil
 554                          )
 555                    (while value
 556                      (setq key (car (car value)))
 557                      ;; (if (find-charset key)
 558                      ;;     (setq cal (cons key cal))
 559                      (setq al (cons key al))
 560                      ;; )
 561                      (setq value (cdr value)))
 562                    (insert-char-attributes ret
 563                                            readable
 564                                            (or al 'none) ; cal
 565                                            nil 'for-sub-node))
 566                (insert (prin1-to-string value)))
 567              (insert ")")
 568              (insert line-breaking))
 569             ((consp value)
 570              (insert (format "(%-18s " name))
 571              (setq lbs (concat "\n" (make-string (current-column) ?\ )))
 572              (while (consp value)
 573                (setq cell (car value))
 574                (if (and (consp cell)
 575                         (consp (car cell))
 576                         (setq ret (condition-case nil
 577                                       (find-char cell)
 578                                     (error nil)))
 579                         )
 580                    (progn
 581                      (setq rest cell
 582                            al nil
 583                            ;; cal nil
 584                            )
 585                      (while rest
 586                        (setq key (car (car rest)))
 587                        ;; (if (find-charset key)
 588                        ;;     (setq cal (cons key cal))
 589                        (setq al (cons key al))
 590                        ;; )
 591                        (setq rest (cdr rest)))
 592                      (if separator
 593                          (insert lbs))
 594                      (insert-char-attributes ret
 595                                              readable
 596                                              al ; cal
 597                                              nil 'for-sub-node)
 598                      (setq separator lbs))
 599                  (if separator
 600                      (insert separator))
 601                  (insert (prin1-to-string cell))
 602                  (setq separator " "))
 603                (setq value (cdr value)))
 604              (insert ")")
 605              (insert line-breaking))
 606             (t
 607              (insert (format "(%-18s . %S)%s"
 608                              name value
 609                              line-breaking))))
 610       (setq alist (cdr alist))))
 611   (insert ")"))
 612
 613 (defun char-db-insert-char-reference (plist &optional readable column)
 614   (unless column
 615     (setq column (current-column)))
 616   (let ((line-breaking
 617          (concat "\n" (make-string (1+ column) ?\ )))
 618         (separator "")
 619         name value)
 620     (insert "(")
 621     (while plist
 622       (setq name (pop plist))
 623       (setq value (pop plist))
 624       (cond ((eq name :char)
 625              (insert separator)
 626              (insert ":char\t")
 627              (cond ((numberp value)
 628                     (setq value (decode-char '=ucs value)))
 629                    ;; ((consp value)
 630                    ;;  (setq value (or (find-char value)
 631                    ;;                  value)))
 632                    )
 633              (char-db-insert-char-spec value readable)
 634              (insert line-breaking)
 635              (setq separator ""))
 636             ((eq name :radical)
 637              (insert (format "%s%s\t%d ; %c%s"
 638                              separator
 639                              name value
 640                              (ideographic-radical value)
 641                              line-breaking))
 642              (setq separator ""))
 643             (t
 644              (insert (format "%s%s\t%S" separator name value))
 645              (setq separator line-breaking)))
 646       ))
 647   (insert ")"))
 648
 649 (defun char-db-decode-isolated-char (ccs code-point)
 650   (let (ret)
 651     (setq ret
 652           (cond ((eq ccs 'arabic-iso8859-6)
 653                  (decode-char ccs code-point))
 654                 ;; ((eq ccs '=gt)
 655                 ;;  (decode-builtin-char '==gt code-point))
 656                 ((and (memq ccs '(=gt-pj-1
 657                                   =gt-pj-2
 658                                   =gt-pj-3
 659                                   =gt-pj-4
 660                                   =gt-pj-5
 661                                   =gt-pj-6
 662                                   =gt-pj-7
 663                                   =gt-pj-8
 664                                   =gt-pj-9
 665                                   =gt-pj-10
 666                                   =gt-pj-11))
 667                       (setq ret (decode-char ccs code-point))
 668                       (setq ret (encode-char ret '=gt 'defined-only)))
 669                  (decode-builtin-char '=gt ret))
 670                 (t
 671                  (decode-builtin-char ccs code-point))))
 672     (cond ((null ret)
 673            (or (decode-char ccs code-point)
 674                (define-char (list (cons ccs code-point)))))
 675           ((and (<= 0 (char-int ret))
 676                 (<= (char-int ret) #x1F))
 677            (decode-char '=ucs (+ #x2400 (char-int ret))))
 678           ((= (char-int ret) #x7F)
 679            ?\u2421)
 680           (t ret))))
 681
 682 (defvar char-db-convert-obsolete-format t)
 683
 684 (defun char-db-insert-ccs-feature (name value line-breaking)
 685   (cond
 686    ((integerp value)
 687     (insert
 688      (format
 689       (cond
 690        ((memq name '(=>iwds-1
 691                      ==shinjigen
 692                      =shinjigen
 693                      =shinjigen@1ed ==shinjigen@1ed
 694                      =shinjigen@rev ==shinjigen@rev
 695                      =shinjigen/+p@rev ==shinjigen/+p@rev
 696                      ===daikanwa/ho ==daikanwa/ho
 697                      =daikanwa/ho =>>daikanwa/ho =>daikanwa/ho))
 698         "(%-18s .  %04d)\t; %c")
 699        ((eq name '=shinjigen@1ed/24pr)
 700         "(%-18s . %04d)\t; %c")
 701        ((or
 702          (memq name
 703                '(===daikanwa
 704                  ==daikanwa =daikanwa =>>daikanwa =>daikanwa
 705                  =daikanwa@rev1 =daikanwa@rev2
 706                  =daikanwa/+p ==daikanwa/+p ===daikanwa/+p
 707                  =>>daikanwa/+p
 708                  =daikanwa/+2p =>>daikanwa/+2p
 709                  =gt ==gt ===gt
 710                  =>>gt =+>gt =>gt
 711                  =gt-k ==gt-k ===gt-k
 712                  =>>gt-k =>gt-k
 713                  =adobe-japan1-0 ==adobe-japan1-0 ===adobe-japan1-0
 714                  =adobe-japan1-1 ==adobe-japan1-1 ===adobe-japan1-1
 715                  =adobe-japan1-2 ==adobe-japan1-2 ===adobe-japan1-2
 716                  =adobe-japan1-3 ==adobe-japan1-3 ===adobe-japan1-3
 717                  =adobe-japan1-4 ==adobe-japan1-4 ===adobe-japan1-4
 718                  =adobe-japan1-5 ==adobe-japan1-5 ===adobe-japan1-5
 719                  =adobe-japan1-6 ==adobe-japan1-6 ===adobe-japan1-6
 720                  =>>adobe-japan1-0 =+>adobe-japan1-0
 721                  =>>adobe-japan1-1 =+>adobe-japan1-1
 722                  =>>adobe-japan1-2 =+>adobe-japan1-2
 723                  =>>adobe-japan1-3 =+>adobe-japan1-3
 724                  =>>adobe-japan1-4 =+>adobe-japan1-4
 725                  =>>adobe-japan1-5 =+>adobe-japan1-5
 726                  =>>adobe-japan1-6 =+>adobe-japan1-6
 727                  =>cbeta =cbeta =>>cbeta ==cbeta ===cbeta
 728                  =zinbun-oracle =>zinbun-oracle
 729                  ===hng-jou ===hng-keg ===hng-dng ===hng-mam
 730                  ===hng-drt ===hng-kgk ===hng-myz ===hng-kda
 731                  ===hng-khi ===hng-khm ===hng-hok ===hng-kyd ===hng-sok
 732                  ===hng-yhk ===hng-kak ===hng-kar ===hng-kae
 733                  ===hng-sys ===hng-tsu ===hng-tzj
 734                  ===hng-hos ===hng-nak ===hng-jhk
 735                  ===hng-hod ===hng-gok ===hng-ink ===hng-nto
 736                  ===hng-nkm ===hng-k24 ===hng-nkk
 737                  ===hng-kcc ===hng-kcj ===hng-kbk ===hng-sik
 738                  ===hng-skk ===hng-kyu ===hng-ksk ===hng-wan
 739                  ===hng-okd ===hng-wad ===hng-kmi
 740                  ===hng-zkd ===hng-doh ===hng-jyu
 741                  ===hng-tzs ===hng-kss ===hng-kyo
 742                  ===hng-smk))
 743          ;; (string-match "^=adobe-" (symbol-name name))
 744          )
 745         "(%-18s . %05d)\t; %c")
 746        ((memq name '(=hanyo-denshi/ks
 747                      ==hanyo-denshi/ks ===hanyo-denshi/ks
 748                      =>>hanyo-denshi/ks
 749                      =koseki ==koseki
 750                      =mj ==mj ===mj =>>mj =>mj
 751                      =zihai mojikyo))
 752         "(%-18s . %06d)\t; %c")
 753        ((memq name '(=hanyo-denshi/tk ==hanyo-denshi/tk))
 754         "(%-18s . %08d)\t; %c")
 755        ((>= (charset-dimension name) 2)
 756         "(%-18s . #x%04X)\t; %c")
 757        (t
 758         "(%-18s . #x%02X)\t; %c"))
 759       name
 760       (if (= (charset-iso-graphic-plane name) 1)
 761           (logior value
 762                   (cond ((= (charset-dimension name) 1)
 763                          #x80)
 764                         ((= (charset-dimension name) 2)
 765                          #x8080)
 766                         ((= (charset-dimension name) 3)
 767                          #x808080)
 768                         (t 0)))
 769         value)
 770       (char-db-decode-isolated-char name value)))
 771     (if (and (= (charset-chars name) 94)
 772              (= (charset-dimension name) 2))
 773         (insert (format " [%02d-%02d]"
 774                         (- (lsh value -8) 32)
 775                         (- (logand value 255) 32))))
 776     )
 777    (t
 778     (insert (format "(%-18s . %s)" name value))
 779     ))
 780   (insert line-breaking))
 781
 782 (defun char-db-insert-relation-feature (char name value line-breaking
 783                                              ccss readable)
 784   (insert (format "(%-18s%s " name line-breaking))
 785   (let ((lbs (concat "\n" (make-string (current-column) ?\ )))
 786         separator cell sources required-features
 787         ret)
 788     (while (consp value)
 789       (setq cell (car value))
 790       (if (integerp cell)
 791           (setq cell (decode-char '=ucs cell)))
 792       (cond
 793        ((eq name '->subsumptive)
 794         (when (or (not (some (lambda (atr)
 795                                (get-char-attribute cell atr))
 796                              char-db-ignored-attributes))
 797                   (some (lambda (ccs)
 798                           (encode-char cell ccs 'defined-only))
 799                         ccss))
 800           (if separator
 801               (insert lbs))
 802           (let ((char-db-ignored-attributes
 803                  (cons '<-subsumptive
 804                        char-db-ignored-attributes)))
 805             (insert-char-attributes cell readable nil nil 'for-sub-node))
 806           (setq separator lbs))
 807         )
 808        ((characterp cell)
 809         (setq sources
 810               (get-char-attribute
 811                char (intern (format "%s*sources" name))))
 812         (setq required-features nil)
 813         (dolist (source sources)
 814           (cond
 815            ((memq source '(JP
 816                            JP/Jouyou
 817                            shinjigen shinjigen@1ed shinjigen@rev))
 818             (setq required-features
 819                   (union required-features
 820                          '(=jis-x0208
 821                            =jis-x0208@1990
 822                            =jis-x0213-1@2000
 823                            =jis-x0213-1@2004
 824                            =jis-x0213-2
 825                            =jis-x0212
 826                            =jis-x0208@1983
 827                            =jis-x0208@1978
 828                            =shinjigen
 829                            =shinjigen@1ed
 830                            =shinjigen@rev
 831                            =shinjigen/+p@rev))))
 832            ((eq source 'CN)
 833             (setq required-features
 834                   (union required-features
 835                          '(=gb2312
 836                            =gb12345
 837                            =iso-ir165)))))
 838           (cond
 839            ((find-charset (setq ret (intern (format "=%s" source))))
 840             (setq required-features
 841                   (cons ret required-features)))
 842            (t (setq required-features
 843                     (cons source required-features)))))
 844         (cond ((string-match "@JP" (symbol-name name))
 845                (setq required-features
 846                      (union required-features
 847                             '(=jis-x0208
 848                               =jis-x0208@1990
 849                               =jis-x0213-1-2000
 850                               =jis-x0213-2-2000
 851                               =jis-x0212
 852                               =jis-x0208@1983
 853                               =jis-x0208@1978))))
 854               ((string-match "@CN" (symbol-name name))
 855                (setq required-features
 856                      (union required-features
 857                             '(=gb2312
 858                               =gb12345
 859                               =iso-ir165)))))
 860         (if separator
 861             (insert lbs))
 862         (if readable
 863             (insert (format "%S" cell))
 864           (char-db-insert-char-spec cell readable
 865                                     nil
 866                                     required-features))
 867         (setq separator lbs))
 868        ((consp cell)
 869         (if separator
 870             (insert lbs))
 871         (if (consp (car cell))
 872             (char-db-insert-char-spec cell readable)
 873           (char-db-insert-char-reference cell readable))
 874         (setq separator lbs))
 875        (t
 876         (if separator
 877             (insert separator))
 878         (insert (prin1-to-string cell))
 879         (setq separator " ")))
 880       (setq value (cdr value)))
 881     (insert ")")
 882     (insert line-breaking)))
 883
 884 (defun insert-char-attributes (char &optional readable attributes column
 885                                     for-sub-node)
 886   (unless column
 887     (setq column (current-column)))
 888   (let (name value ; has-long-ccs-name
 889         rest
 890         radical strokes
 891         (line-breaking
 892          (concat "\n" (make-string (1+ column) ?\ )))
 893         lbs cell separator ret
 894         key al cal
 895         dest-ccss ; sources required-features
 896         ccss)
 897     (let (atr-d)
 898       (setq attributes
 899             (sort (if attributes
 900                       (if (consp attributes)
 901                           (progn
 902                             (dolist (name attributes)
 903                               (unless (memq name char-db-ignored-attributes)
 904                                 (if (find-charset name)
 905                                     (push name ccss))
 906                                 (push name atr-d)))
 907                             atr-d))
 908                     (dolist (name (char-attribute-list))
 909                       (unless (memq name char-db-ignored-attributes)
 910                         (if (find-charset name)
 911                             (push name ccss))
 912                         (push name atr-d)))
 913                     atr-d)
 914                   #'char-attribute-name<)))
 915     (insert "(")
 916     (when (memq '<-subsumptive attributes)
 917       (when (or readable (not for-sub-node))
 918         (when (setq value (get-char-attribute char '<-subsumptive))
 919           (char-db-insert-relation-feature char '<-subsumptive value
 920                                            line-breaking
 921                                            ccss readable)))
 922       (setq attributes (delq '<-subsumptive attributes)))
 923     (when (and (memq '<-denotational attributes)
 924                (setq value (get-char-attribute char '<-denotational)))
 925       (char-db-insert-relation-feature char '<-denotational value
 926                                        line-breaking
 927                                        ccss readable)
 928       (setq attributes (delq '<-denotational attributes)))
 929     (when (and (memq '<-denotational@component attributes)
 930                (setq value (get-char-attribute char '<-denotational@component)))
 931       (char-db-insert-relation-feature char '<-denotational@component value
 932                                        line-breaking
 933                                        ccss readable)
 934       (setq attributes (delq '<-denotational@component attributes)))
 935     (when (and (memq 'name attributes)
 936                (setq value (get-char-attribute char 'name)))
 937       (insert (format
 938                (if (> (+ (current-column) (length value)) 48)
 939                    "(name . %S)%s"
 940                  "(name               . %S)%s")
 941                value line-breaking))
 942       (setq attributes (delq 'name attributes))
 943       )
 944     (when (and (memq 'name* attributes)
 945                (setq value (get-char-attribute char 'name*)))
 946       (insert (format
 947                (if (> (+ (current-column) (length value)) 48)
 948                    "(name* . %S)%s"
 949                  "(name*              . %S)%s")
 950                value line-breaking))
 951       (setq attributes (delq 'name* attributes))
 952       )
 953     (when (and (memq 'script attributes)
 954                (setq value (get-char-attribute char 'script)))
 955       (insert (format "(script\t\t%s)%s"
 956                       (mapconcat (function prin1-to-string)
 957                                  value " ")
 958                       line-breaking))
 959       (setq attributes (delq 'script attributes))
 960       )
 961     (dolist (name '(=>ucs =>ucs*))
 962       (when (and (memq name attributes)
 963                  (setq value (get-char-attribute char name)))
 964         (insert (format "(%-18s . #x%04X)\t; %c%s"
 965                         name value (decode-char '=ucs value)
 966                         line-breaking))
 967         (setq attributes (delq name attributes))))
 968     (dolist (name '(=>ucs@gb =>ucs@big5))
 969       (when (and (memq name attributes)
 970                  (setq value (get-char-attribute char name)))
 971         (insert (format "(%-18s . #x%04X)\t; %c%s"
 972                         name value
 973                         (decode-char (intern
 974                                       (concat "="
 975                                               (substring
 976                                                (symbol-name name) 2)))
 977                                      value)
 978                         line-breaking))
 979         (setq attributes (delq name attributes))
 980         ))
 981     ;; (dolist (name '(=>daikanwa))
 982     ;;   (when (and (memq name attributes)
 983     ;;              (setq value (get-char-attribute char name)))
 984     ;;     (insert
 985     ;;      (if (integerp value)
 986     ;;          (format "(%-18s . %05d)\t; %c%s"
 987     ;;                  name value (decode-char '=daikanwa value)
 988     ;;                  line-breaking)
 989     ;;        (format "(%-18s %s)\t; %c%s"
 990     ;;                name
 991     ;;                (mapconcat (function prin1-to-string)
 992     ;;                           value " ")
 993     ;;                (char-representative-of-daikanwa char)
 994     ;;                line-breaking)))
 995     ;;     (setq attributes (delq name attributes))))
 996     (when (and (memq 'general-category attributes)
 997                (setq value (get-char-attribute char 'general-category)))
 998       (insert (format
 999                "(general-category\t%s) ; %s%s"
1000                (mapconcat (lambda (cell)
1001                             (format "%S" cell))
1002                           value " ")
1003                (cond ((rassoc value unidata-normative-category-alist)
1004                       "Normative Category")
1005                      ((rassoc value unidata-informative-category-alist)
1006                       "Informative Category")
1007                      (t
1008                       "Unknown Category"))
1009                line-breaking))
1010       (setq attributes (delq 'general-category attributes))
1011       )
1012     (when (and (memq 'bidi-category attributes)
1013                (setq value (get-char-attribute char 'bidi-category)))
1014       (insert (format "(bidi-category\t. %S)%s"
1015                       value
1016                       line-breaking))
1017       (setq attributes (delq 'bidi-category attributes))
1018       )
1019     (unless (or (not (memq 'mirrored attributes))
1020                 (eq (setq value (get-char-attribute char 'mirrored 'empty))
1021                     'empty))
1022       (insert (format "(mirrored\t\t. %S)%s"
1023                       value
1024                       line-breaking))
1025       (setq attributes (delq 'mirrored attributes))
1026       )
1027     (cond
1028      ((and (memq 'decimal-digit-value attributes)
1029            (setq value (get-char-attribute char 'decimal-digit-value)))
1030       (insert (format "(decimal-digit-value . %S)%s"
1031                       value
1032                       line-breaking))
1033       (setq attributes (delq 'decimal-digit-value attributes))
1034       (when (and (memq 'digit-value attributes)
1035                  (setq value (get-char-attribute char 'digit-value)))
1036         (insert (format "(digit-value\t . %S)%s"
1037                         value
1038                         line-breaking))
1039         (setq attributes (delq 'digit-value attributes))
1040         )
1041       (when (and (memq 'numeric-value attributes)
1042                  (setq value (get-char-attribute char 'numeric-value)))
1043         (insert (format "(numeric-value\t . %S)%s"
1044                         value
1045                         line-breaking))
1046         (setq attributes (delq 'numeric-value attributes))
1047         )
1048       )
1049      (t
1050       (when (and (memq 'digit-value attributes)
1051                  (setq value (get-char-attribute char 'digit-value)))
1052         (insert (format "(digit-value\t. %S)%s"
1053                         value
1054                         line-breaking))
1055         (setq attributes (delq 'digit-value attributes))
1056         )
1057       (when (and (memq 'numeric-value attributes)
1058                  (setq value (get-char-attribute char 'numeric-value)))
1059         (insert (format "(numeric-value\t. %S)%s"
1060                         value
1061                         line-breaking))
1062         (setq attributes (delq 'numeric-value attributes))
1063         )))
1064     (when (and (memq 'iso-10646-comment attributes)
1065                (setq value (get-char-attribute char 'iso-10646-comment)))
1066       (insert (format "(iso-10646-comment\t. %S)%s"
1067                       value
1068                       line-breaking))
1069       (setq attributes (delq 'iso-10646-comment attributes))
1070       )
1071     (when (and (memq 'morohashi-daikanwa attributes)
1072                (setq value (get-char-attribute char 'morohashi-daikanwa)))
1073       (insert (format "(morohashi-daikanwa\t%s)%s"
1074                       (mapconcat (function prin1-to-string) value " ")
1075                       line-breaking))
1076       (setq attributes (delq 'morohashi-daikanwa attributes))
1077       )
1078     (setq radical nil
1079           strokes nil)
1080     (when (and (memq 'ideographic-radical attributes)
1081                (setq value (get-char-attribute char 'ideographic-radical)))
1082       (setq radical value)
1083       (insert (format "(ideographic-radical . %S)\t; %c%s"
1084                       radical
1085                       (ideographic-radical radical)
1086                       line-breaking))
1087       (setq attributes (delq 'ideographic-radical attributes))
1088       )
1089     (when (and (memq 'shuowen-radical attributes)
1090                (setq value (get-char-attribute char 'shuowen-radical)))
1091       (insert (format "(shuowen-radical\t. %S)\t; %c%s"
1092                       value
1093                       (shuowen-radical value)
1094                       line-breaking))
1095       (setq attributes (delq 'shuowen-radical attributes))
1096       )
1097     (let (key)
1098       (dolist (domain
1099                (append
1100                 char-db-feature-domains
1101                 (let (dest domain)
1102                   (dolist (feature (char-attribute-list))
1103                     (setq feature (symbol-name feature))
1104                     (when (string-match
1105                            "\\(radical\\|strokes\\)@\\([^@*]+\\)\\(\\*\\|$\\)"
1106                            feature)
1107                       (setq domain (intern (match-string 2 feature)))
1108                      (unless (memq domain dest)
1109                        (setq dest (cons domain dest)))))
1110                   (sort dest #'string<))))
1111         (setq key (intern (format "%s@%s" 'ideographic-radical domain)))
1112         (when (and (memq key attributes)
1113                    (setq value (get-char-attribute char key)))
1114           (setq radical value)
1115           (insert (format "(%s . %S)\t; %c%s"
1116                           key
1117                           radical
1118                           (ideographic-radical radical)
1119                           line-breaking))
1120           (setq attributes (delq key attributes))
1121           )
1122         (setq key (intern (format "%s@%s" 'ideographic-strokes domain)))
1123         (when (and (memq key attributes)
1124                    (setq value (get-char-attribute char key)))
1125           (setq strokes value)
1126           (insert (format "(%s . %S)%s"
1127                           key
1128                           strokes
1129                           line-breaking))
1130           (setq attributes (delq key attributes))
1131           )
1132         (setq key (intern (format "%s@%s" 'total-strokes domain)))
1133         (when (and (memq key attributes)
1134                    (setq value (get-char-attribute char key)))
1135           (insert (format "(%s       . %S)%s"
1136                           key
1137                           value
1138                           line-breaking))
1139           (setq attributes (delq key attributes))
1140           )
1141         (dolist (feature '(ideographic-radical
1142                            ideographic-strokes
1143                            total-strokes))
1144           (setq key (intern (format "%s@%s*sources" feature domain)))
1145           (when (and (memq key attributes)
1146                      (setq value (get-char-attribute char key)))
1147             (insert (format "(%s%s" key line-breaking))
1148             (dolist (cell value)
1149               (insert (format " %s" cell)))
1150             (insert ")")
1151             (insert line-breaking)
1152             (setq attributes (delq key attributes))
1153             ))
1154         ))
1155     (when (and (memq 'ideographic-strokes attributes)
1156                (setq value (get-char-attribute char 'ideographic-strokes)))
1157       (setq strokes value)
1158       (insert (format "(ideographic-strokes . %S)%s"
1159                       strokes
1160                       line-breaking))
1161       (setq attributes (delq 'ideographic-strokes attributes))
1162       )
1163     (when (and (memq 'kangxi-radical attributes)
1164                (setq value (get-char-attribute char 'kangxi-radical)))
1165       (unless (eq value radical)
1166         (insert (format "(kangxi-radical\t . %S)\t; %c%s"
1167                         value
1168                         (ideographic-radical value)
1169                         line-breaking))
1170         (or radical
1171             (setq radical value)))
1172       (setq attributes (delq 'kangxi-radical attributes))
1173       )
1174     (when (and (memq 'kangxi-strokes attributes)
1175                (setq value (get-char-attribute char 'kangxi-strokes)))
1176       (unless (eq value strokes)
1177         (insert (format "(kangxi-strokes\t . %S)%s"
1178                         value
1179                         line-breaking))
1180         (or strokes
1181             (setq strokes value)))
1182       (setq attributes (delq 'kangxi-strokes attributes))
1183       )
1184     ;; (when (and (memq 'japanese-radical attributes)
1185     ;;            (setq value (get-char-attribute char 'japanese-radical)))
1186     ;;   (unless (eq value radical)
1187     ;;     (insert (format "(japanese-radical\t . %S)\t; %c%s"
1188     ;;                     value
1189     ;;                     (ideographic-radical value)
1190     ;;                     line-breaking))
1191     ;;     (or radical
1192     ;;         (setq radical value)))
1193     ;;   (setq attributes (delq 'japanese-radical attributes))
1194     ;;   )
1195     (when (and (memq 'japanese-strokes attributes)
1196                (setq value (get-char-attribute char 'japanese-strokes)))
1197       (unless (eq value strokes)
1198         (insert (format "(japanese-strokes\t . %S)%s"
1199                         value
1200                         line-breaking))
1201         (or strokes
1202             (setq strokes value)))
1203       (setq attributes (delq 'japanese-strokes attributes))
1204       )
1205     (when (and (memq 'cns-radical attributes)
1206                (setq value (get-char-attribute char 'cns-radical)))
1207       (insert (format "(cns-radical\t . %S)\t; %c%s"
1208                       value
1209                       (ideographic-radical value)
1210                       line-breaking))
1211       (setq attributes (delq 'cns-radical attributes))
1212       )
1213     (when (and (memq 'cns-strokes attributes)
1214                (setq value (get-char-attribute char 'cns-strokes)))
1215       (unless (eq value strokes)
1216         (insert (format "(cns-strokes\t . %S)%s"
1217                         value
1218                         line-breaking))
1219         (or strokes
1220             (setq strokes value)))
1221       (setq attributes (delq 'cns-strokes attributes))
1222       )
1223     ;; (when (and (memq 'shinjigen-1-radical attributes)
1224     ;;            (setq value (get-char-attribute char 'shinjigen-1-radical)))
1225     ;;   (unless (eq value radical)
1226     ;;     (insert (format "(shinjigen-1-radical . %S)\t; %c%s"
1227     ;;                     value
1228     ;;                     (ideographic-radical value)
1229     ;;                     line-breaking))
1230     ;;     (or radical
1231     ;;         (setq radical value)))
1232     ;;   (setq attributes (delq 'shinjigen-1-radical attributes))
1233     ;;   )
1234     ;; (when (and (memq 'ideographic- attributes)
1235     ;;            (setq value (get-char-attribute char 'ideographic-)))
1236     ;;   (insert "(ideographic-       ")
1237     ;;   (setq lbs (concat "\n" (make-string (current-column) ?\ ))
1238     ;;         separator nil)
1239     ;;   (while (consp value)
1240     ;;     (setq cell (car value))
1241     ;;     (if (integerp cell)
1242     ;;         (setq cell (decode-char '=ucs cell)))
1243     ;;     (cond ((characterp cell)
1244     ;;            (if separator
1245     ;;                (insert lbs))
1246     ;;            (if readable
1247     ;;                (insert (format "%S" cell))
1248     ;;              (char-db-insert-char-spec cell readable))
1249     ;;            (setq separator lbs))
1250     ;;           ((consp cell)
1251     ;;            (if separator
1252     ;;                (insert lbs))
1253     ;;            (if (consp (car cell))
1254     ;;                (char-db-insert-char-spec cell readable)
1255     ;;              (char-db-insert-char-reference cell readable))
1256     ;;            (setq separator lbs))
1257     ;;           (t
1258     ;;            (if separator
1259     ;;                (insert separator))
1260     ;;            (insert (prin1-to-string cell))
1261     ;;            (setq separator " ")))
1262     ;;     (setq value (cdr value)))
1263     ;;   (insert ")")
1264     ;;   (insert line-breaking)
1265     ;;   (setq attributes (delq 'ideographic- attributes)))
1266     (when (and (memq 'total-strokes attributes)
1267                (setq value (get-char-attribute char 'total-strokes)))
1268       (insert (format "(total-strokes       . %S)%s"
1269                       value
1270                       line-breaking))
1271       (setq attributes (delq 'total-strokes attributes))
1272       )
1273     ;; (when (and (memq '->ideograph attributes)
1274     ;;            (setq value (get-char-attribute char '->ideograph)))
1275     ;;   (insert (format "(->ideograph\t%s)%s"
1276     ;;                   (mapconcat (lambda (code)
1277     ;;                                (cond ((symbolp code)
1278     ;;                                       (symbol-name code))
1279     ;;                                      ((integerp code)
1280     ;;                                       (format "#x%04X" code))
1281     ;;                                      (t
1282     ;;                                       (format "%s %S"
1283     ;;                                               line-breaking code))))
1284     ;;                              value " ")
1285     ;;                   line-breaking))
1286     ;;   (setq attributes (delq '->ideograph attributes))
1287     ;;   )
1288     ;; (when (and (memq '->decomposition attributes)
1289     ;;            (setq value (get-char-attribute char '->decomposition)))
1290     ;;   (insert (format "(->decomposition\t%s)%s"
1291     ;;                   (mapconcat (lambda (code)
1292     ;;                                (cond ((symbolp code)
1293     ;;                                       (symbol-name code))
1294     ;;                                      ((characterp code)
1295     ;;                                       (if readable
1296     ;;                                           (format "%S" code)
1297     ;;                                         (format "#x%04X"
1298     ;;                                                 (char-int code))
1299     ;;                                         ))
1300     ;;                                      ((integerp code)
1301     ;;                                       (format "#x%04X" code))
1302     ;;                                      (t
1303     ;;                                       (format "%s%S" line-breaking code))))
1304     ;;                              value " ")
1305     ;;                   line-breaking))
1306     ;;   (setq attributes (delq '->decomposition attributes))
1307     ;;   )
1308     (if (equal (get-char-attribute char '->titlecase)
1309                (get-char-attribute char '->uppercase))
1310         (setq attributes (delq '->titlecase attributes)))
1311     (when (and (memq '->mojikyo attributes)
1312                (setq value (get-char-attribute char '->mojikyo)))
1313       (insert (format "(->mojikyo\t\t. %06d)\t; %c%s"
1314                       value (decode-char 'mojikyo value)
1315                       line-breaking))
1316       (setq attributes (delq '->mojikyo attributes))
1317       )
1318     ;; (when (and (memq 'hanyu-dazidian-vol attributes)
1319     ;;            (setq value (get-char-attribute char 'hanyu-dazidian-vol)))
1320     ;;   (insert (format "(hanyu-dazidian-vol  . %d)%s"
1321     ;;                   value line-breaking))
1322     ;;   (setq attributes (delq 'hanyu-dazidian-vol attributes))
1323     ;;   )
1324     ;; (when (and (memq 'hanyu-dazidian-page attributes)
1325     ;;            (setq value (get-char-attribute char 'hanyu-dazidian-page)))
1326     ;;   (insert (format "(hanyu-dazidian-page . %d)%s"
1327     ;;                   value line-breaking))
1328     ;;   (setq attributes (delq 'hanyu-dazidian-page attributes))
1329     ;;   )
1330     ;; (when (and (memq 'hanyu-dazidian-char attributes)
1331     ;;            (setq value (get-char-attribute char 'hanyu-dazidian-char)))
1332     ;;   (insert (format "(hanyu-dazidian-char . %d)%s"
1333     ;;                   value line-breaking))
1334     ;;   (setq attributes (delq 'hanyu-dazidian-char attributes))
1335     ;;   )
1336     (unless readable
1337       (dolist (ignored '(composition
1338                          ->denotational <-subsumptive ->ucs-unified
1339                          ->ideographic-component-forms))
1340         (setq attributes (delq ignored attributes))))
1341     (while attributes
1342       (setq name (car attributes))
1343       (unless (eq (setq value (get-char-attribute char name 'value-is-empty))
1344                   'value-is-empty)
1345         (cond ((setq ret (find-charset name))
1346                (setq name (charset-name ret))
1347                (when (not (memq name dest-ccss))
1348                  (setq dest-ccss (cons name dest-ccss))
1349                  (char-db-insert-ccs-feature name value line-breaking))
1350                )
1351               ((string-match "^=>ucs@" (symbol-name name))
1352                (insert (format "(%-18s . #x%04X)\t; %c%s"
1353                                name value (decode-char '=ucs value)
1354                                line-breaking))
1355                )
1356               ((eq name 'jisx0208-1978/4X)
1357                (insert (format "(%-18s . #x%04X)%s"
1358                                name value
1359                                line-breaking))
1360                )
1361               ((and
1362                 (not readable)
1363                 (not (eq name '->subsumptive))
1364                 (not (eq name '->uppercase))
1365                 (not (eq name '->lowercase))
1366                 (not (eq name '->titlecase))
1367                 (not (eq name '->canonical))
1368                 (not (eq name '->Bopomofo))
1369                 (not (eq name '->mistakable))
1370                 (not (eq name '->ideographic-variants))
1371                 (null (get-char-attribute
1372                        char (intern (format "%s*sources" name))))
1373                 (not (string-match "\\*sources$" (symbol-name name)))
1374                 (null (get-char-attribute
1375                        char (intern (format "%s*note" name))))
1376                 (not (string-match "\\*note$" (symbol-name name)))
1377                 (or (eq name '<-identical)
1378                     (eq name '<-uppercase)
1379                     (eq name '<-lowercase)
1380                     (eq name '<-titlecase)
1381                     (eq name '<-canonical)
1382                     (eq name '<-ideographic-variants)
1383                     ;; (eq name '<-synonyms)
1384                     (string-match "^<-synonyms" (symbol-name name))
1385                     (eq name '<-mistakable)
1386                     (when (string-match "^->" (symbol-name name))
1387                       (cond
1388                        ((string-match "^->fullwidth" (symbol-name name))
1389                         (not (and (consp value)
1390                                   (characterp (car value))
1391                                   (encode-char
1392                                    (car value) '=ucs 'defined-only)))
1393                         )
1394                        (t)))
1395                     ))
1396                )
1397               ((or (eq name 'ideographic-structure)
1398                    (eq name 'ideographic-combination)
1399                    ;; (eq name 'ideographic-)
1400                    (eq name '=decomposition)
1401                    (char-feature-base-name= '=decomposition name)
1402                    (char-feature-base-name= '=>decomposition name)
1403                    ;; (string-match "^=>*decomposition\\(@[^*]+\\)?$"
1404                    ;;               (symbol-name name))
1405                    (string-match "^\\(->\\|<-\\)[^*]*$" (symbol-name name))
1406                    (string-match "^\\(->\\|<-\\)[^*]*\\*sources$"
1407                                  (symbol-name name))
1408                    )
1409                (char-db-insert-relation-feature char name value
1410                                                 line-breaking
1411                                                 ccss readable))
1412               ;; ((memq name '(ideograph=
1413               ;;               original-ideograph-of
1414               ;;               ancient-ideograph-of
1415               ;;               vulgar-ideograph-of
1416               ;;               wrong-ideograph-of
1417               ;;               ;; simplified-ideograph-of
1418               ;;               ideographic-variants
1419               ;;               ;; ideographic-different-form-of
1420               ;;               ))
1421               ;;  (insert (format "(%-18s%s " name line-breaking))
1422               ;;  (setq lbs (concat "\n" (make-string (current-column) ?\ ))
1423               ;;        separator nil)
1424               ;;  (while (consp value)
1425               ;;    (setq cell (car value))
1426               ;;    (if (and (consp cell)
1427               ;;             (consp (car cell)))
1428               ;;        (progn
1429               ;;          (if separator
1430               ;;              (insert lbs))
1431               ;;          (char-db-insert-alist cell readable)
1432               ;;          (setq separator lbs))
1433               ;;      (if separator
1434               ;;          (insert separator))
1435               ;;      (insert (prin1-to-string cell))
1436               ;;      (setq separator " "))
1437               ;;    (setq value (cdr value)))
1438               ;;  (insert ")")
1439               ;;  (insert line-breaking))
1440               ((consp value)
1441                (insert (format "(%-18s " name))
1442                (setq lbs (concat "\n" (make-string (current-column) ?\ ))
1443                      separator nil)
1444                (while (consp value)
1445                  (setq cell (car value))
1446                  (if (and (consp cell)
1447                           (consp (car cell))
1448                           (setq ret (condition-case nil
1449                                         (find-char cell)
1450                                       (error nil))))
1451                      (progn
1452                        (setq rest cell
1453                              al nil
1454                              cal nil)
1455                        (while rest
1456                          (setq key (car (car rest)))
1457                          (if (find-charset key)
1458                              (setq cal (cons key cal))
1459                            (setq al (cons key al)))
1460                          (setq rest (cdr rest)))
1461                        (if separator
1462                            (insert lbs))
1463                        (insert-char-attributes ret
1464                                                readable
1465                                                al ; cal
1466                                                nil 'for-sub-node)
1467                        (setq separator lbs))
1468                    (setq ret (prin1-to-string cell))
1469                    (if separator
1470                        (if (< (+ (current-column)
1471                                  (length ret)
1472                                  (length separator))
1473                               76)
1474                            (insert separator)
1475                          (insert lbs)))
1476                    (insert ret)
1477                    (setq separator " "))
1478                  (setq value (cdr value)))
1479                (insert ")")
1480                (insert line-breaking))
1481               (t
1482                (insert (format "(%-18s" name))
1483                (setq ret (prin1-to-string value))
1484                (unless (< (+ (current-column)
1485                              (length ret)
1486                              3)
1487                           76)
1488                  (insert line-breaking))
1489                (insert " . " ret ")" line-breaking)
1490                ;; (insert (format "(%-18s . %S)%s"
1491                ;;                 name value
1492                ;;                 line-breaking))
1493                )
1494               ))
1495       (setq attributes (cdr attributes)))
1496     (insert ")")))
1497
1498 (defun insert-char-data (char &optional readable
1499                               attributes)
1500   (save-restriction
1501     (narrow-to-region (point)(point))
1502     (insert "(define-char
1503   '")
1504     (insert-char-attributes char readable attributes)
1505     (insert ")\n")
1506     (goto-char (point-min))
1507     (while (re-search-forward "[ \t]+$" nil t)
1508       (replace-match ""))
1509     ;; from tabify.
1510     (goto-char (point-min))
1511     (while (re-search-forward "[ \t][ \t][ \t]*" nil t)
1512       (let ((column (current-column))
1513             (indent-tabs-mode t))
1514         (delete-region (match-beginning 0) (point))
1515         (indent-to column)))
1516     (goto-char (point-max))
1517     ;; (tabify (point-min)(point-max))
1518     ))
1519
1520 (defun insert-char-data-with-variant (char &optional printable
1521                                            no-ucs-unified
1522                                            script excluded-script)
1523   (insert-char-data char printable)
1524   (let ((variants (char-variants char))
1525         rest
1526         variant vs ret)
1527     (setq variants (sort variants #'<))
1528     (setq rest variants)
1529     (setq variants (cons char variants))
1530     (while rest
1531       (setq variant (car rest))
1532       (unless (get-char-attribute variant '<-subsumptive)
1533         (if (and (or (null script)
1534                      (null (setq vs (get-char-attribute variant 'script)))
1535                      (memq script vs))
1536                  (or (null excluded-script)
1537                      (null (setq vs (get-char-attribute variant 'script)))
1538                      (not (memq excluded-script vs))))
1539             (unless (and no-ucs-unified (get-char-attribute variant '=ucs))
1540               (insert-char-data variant printable)
1541               (if (setq ret (char-variants variant))
1542                   (while ret
1543                     (or (memq (car ret) variants)
1544                         ;; (get-char-attribute (car ret) '<-subsumptive)
1545                         (setq rest (nconc rest (list (car ret)))))
1546                     (setq ret (cdr ret)))))))
1547       (setq rest (cdr rest)))))
1548
1549 (defun insert-char-range-data (min max &optional script excluded-script)
1550   (let ((code min)
1551         char)
1552     (while (<= code max)
1553       (setq char (decode-char '=ucs code))
1554       (if (encode-char char '=ucs 'defined-only)
1555           (insert-char-data-with-variant char nil 'no-ucs-unified
1556                                          script excluded-script))
1557       (setq code (1+ code)))))
1558
1559 (defun write-char-range-data-to-file (min max file
1560                                           &optional script excluded-script)
1561   (let ((coding-system-for-write char-db-file-coding-system))
1562     (with-temp-buffer
1563       (insert (format ";; -*- coding: %s -*-\n"
1564                       char-db-file-coding-system))
1565       (insert-char-range-data min max script excluded-script)
1566       (write-region (point-min)(point-max) file))))
1567
1568 (defvar what-character-original-window-configuration)
1569
1570 ;;;###autoload
1571 (defun what-char-definition (char)
1572   (interactive (list (char-after)))
1573   (let ((buf (get-buffer-create "*Character Description*"))
1574         (the-buf (current-buffer))
1575         (win-conf (current-window-configuration)))
1576     (pop-to-buffer buf)
1577     (make-local-variable 'what-character-original-window-configuration)
1578     (setq what-character-original-window-configuration win-conf)
1579     (setq buffer-read-only nil)
1580     (erase-buffer)
1581     (condition-case err
1582         (progn
1583           (insert-char-data-with-variant char 'printable)
1584           (unless (char-attribute-alist char)
1585             (insert (format ";; = %c\n"
1586                             (let* ((rest (split-char char))
1587                                    (ccs (pop rest))
1588                                    (code (pop rest)))
1589                               (while rest
1590                                 (setq code (logior (lsh code 8)
1591                                                    (pop rest))))
1592                               (decode-char ccs code)))))
1593           ;; (char-db-update-comment)
1594           (set-buffer-modified-p nil)
1595           (view-mode the-buf (lambda (buf)
1596                                (set-window-configuration
1597                                 what-character-original-window-configuration)
1598                                ))
1599           (goto-char (point-min)))
1600       (error (progn
1601                (set-window-configuration
1602                 what-character-original-window-configuration)
1603                (signal (car err) (cdr err)))))))
1604
1605
1606 ;;; @ end
1607 ;;;
1608
1609 (provide 'char-db-util)
1610
1611 ;;; char-db-util.el ends here