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 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-011
 350     =>ucs-itaiji-001
 351     =>ucs-itaiji-002
 352     =>ucs-itaiji-003
 353     =>ucs-itaiji-004
 354     =>ucs-itaiji-005
 355     =>ucs-itaiji-006
 356     =>ucs-itaiji-007
 357     =big5-cdp-var-001
 358     =big5-cdp-var-002
 359     =big5-cdp-var-003
 360     =big5-cdp-var-004
 361     =big5-cdp-var-005
 362     =big5-cdp-itaiji-001
 363     =big5-cdp-itaiji-002
 364     =big5-cdp-itaiji-003
 365     =>big5-cdp-itaiji-001
 366     =>ucs@iwds-1
 367     =>ucs@cognate
 368     =>ucs@component
 369     ==ucs@gb
 370     =ucs@gb
 371     ==ucs-var-002
 372     =ucs@JP/hanazono))
 373
 374
 375 ;;; @ char-db formatters
 376 ;;;
 377
 378 (defun char-db-make-char-spec (char)
 379   (let (ret char-spec)
 380     (cond ((characterp char)
 381            (cond ((and (setq ret (encode-char char '=ucs 'defined-only))
 382                        (not (and (<= #xE000 ret)(<= ret #xF8FF))))
 383                   (setq char-spec (list (cons '=ucs ret)))
 384                   (cond ((setq ret (get-char-attribute char 'name))
 385                          (setq char-spec (cons (cons 'name ret) char-spec))
 386                          )
 387                         ((setq ret (get-char-attribute char 'name*))
 388                          (setq char-spec (cons (cons 'name* ret) char-spec))
 389                          ))
 390                   )
 391                  ((encode-char char '=mj 'defined-only)
 392                   (setq char-spec nil)
 393                   (dolist (ccs (charset-list))
 394                     (if (and (or (eq ccs '=mj)
 395                                  ;; (eq (charset-property ccs 'iso-ir) 177)
 396                                  (string-match "=ucs@" (symbol-name ccs))
 397                                  )
 398                              (setq ccs (charset-name ccs))
 399                              (null (assq ccs char-spec))
 400                              (setq ret (encode-char char ccs 'defined-only)))
 401                         (setq char-spec (cons (cons ccs ret) char-spec))))
 402                   )
 403                  ((encode-char char '==mj 'defined-only)
 404                   (setq char-spec nil)
 405                   (dolist (ccs (charset-list))
 406                     (if (and (or (eq ccs '==mj)
 407                                  ;; (eq (charset-property ccs 'iso-ir) 177)
 408                                  (string-match "=ucs@" (symbol-name ccs))
 409                                  )
 410                              (setq ccs (charset-name ccs))
 411                              (null (assq ccs char-spec))
 412                              (setq ret (encode-char char ccs 'defined-only)))
 413                         (setq char-spec (cons (cons ccs ret) char-spec))))
 414                   )
 415                  ((encode-char char '=adobe-japan1 'defined-only)
 416                   (setq char-spec nil)
 417                   (dolist (ccs (charset-list))
 418                     (if (and (or (memq ccs
 419                                        '(=adobe-japan1-0
 420                                          =adobe-japan1-1
 421                                          =adobe-japan1-2
 422                                          =adobe-japan1-3
 423                                          =adobe-japan1-4
 424                                          =adobe-japan1-5
 425                                          =adobe-japan1-6
 426                                          ))
 427                                  ;; (eq (charset-property ccs 'iso-ir) 177)
 428                                  (string-match "=ucs@" (symbol-name ccs))
 429                                  )
 430                              (setq ccs (charset-name ccs))
 431                              (null (assq ccs char-spec))
 432                              (setq ret (encode-char char ccs 'defined-only)))
 433                         (setq char-spec (cons (cons ccs ret) char-spec))))
 434                   )
 435                  ((encode-char char '==adobe-japan1 'defined-only)
 436                   (setq char-spec nil)
 437                   (dolist (ccs (charset-list))
 438                     (if (and (or (memq ccs
 439                                        '(==adobe-japan1-0
 440                                          ==adobe-japan1-1
 441                                          ==adobe-japan1-2
 442                                          ==adobe-japan1-3
 443                                          ==adobe-japan1-4
 444                                          ==adobe-japan1-5
 445                                          ==adobe-japan1-6
 446                                          ))
 447                                  ;; (eq (charset-property ccs 'iso-ir) 177)
 448                                  (string-match "=ucs@" (symbol-name ccs))
 449                                  )
 450                              (setq ccs (charset-name ccs))
 451                              (null (assq ccs char-spec))
 452                              (setq ret (encode-char char ccs 'defined-only)))
 453                         (setq char-spec (cons (cons ccs ret) char-spec))))
 454                   )
 455                  ((setq ret
 456                         (catch 'tag
 457                           (let ((rest char-db-coded-charset-priority-list)
 458                                 ccs)
 459                             (while rest
 460                               (setq ccs (charset-name
 461                                          (find-charset (car rest))))
 462                               (if (setq ret
 463                                         (encode-char char ccs
 464                                                      'defined-only))
 465                                   (throw 'tag (cons ccs ret)))
 466                               (setq rest (cdr rest))))))
 467                   (setq char-spec (list ret))
 468                   (dolist (ccs (delq (car ret) (charset-list)))
 469                     (if (and (or (charset-iso-final-char ccs)
 470                                  (memq ccs
 471                                        '(=daikanwa
 472                                          =daikanwa@rev2
 473                                          ;; =gt-k
 474                                          =jis-x0208@1997
 475                                          ))
 476                                  (eq (charset-property ccs 'iso-ir) 177)
 477                                  ;; (string-match "=ucs@" (symbol-name ccs))
 478                                  )
 479                              (setq ccs (charset-name ccs))
 480                              (null (assq ccs char-spec))
 481                              (setq ret (encode-char char ccs 'defined-only)))
 482                         (setq char-spec (cons (cons ccs ret) char-spec))))
 483                   (if (null char-spec)
 484                       (setq char-spec (split-char char)))
 485                   (cond ((setq ret (get-char-attribute char 'name))
 486                          (setq char-spec (cons (cons 'name ret) char-spec))
 487                          )
 488                         ((setq ret (get-char-attribute char 'name*))
 489                          (setq char-spec (cons (cons 'name* ret) char-spec))
 490                          ))
 491                   )
 492                  ((setq ret (get-char-attribute
 493                              char 'ideographic-combination))
 494                   (setq char-spec
 495                         (cons (cons 'ideographic-combination ret)
 496                               char-spec))
 497                   ))
 498            char-spec)
 499           ((consp char)
 500            char))))
 501
 502 (defun char-db-insert-char-spec (char &optional readable column
 503                                       required-features)
 504   (unless column
 505     (setq column (current-column)))
 506   (let (char-spec temp-char)
 507     (setq char-spec (char-db-make-char-spec char))
 508     (unless (or (characterp char) ; char
 509                 (condition-case nil
 510                     (setq char (find-char char-spec))
 511                   (error nil)))
 512       ;; define temporary character
 513       ;;   Current implementation is dirty.
 514       (setq temp-char (define-char (cons '(ideograph-daikanwa . 0)
 515                                          char-spec)))
 516       (remove-char-attribute temp-char 'ideograph-daikanwa)
 517       (setq char temp-char))
 518     (insert-char-attributes char
 519                             readable
 520                             (union (mapcar #'car char-spec)
 521                                    required-features)
 522                             nil 'for-sub-node)
 523     (when temp-char
 524       ;; undefine temporary character
 525       ;;   Current implementation is dirty.
 526       (setq char-spec (char-attribute-alist temp-char))
 527       (while char-spec
 528         (remove-char-attribute temp-char (car (car char-spec)))
 529         (setq char-spec (cdr char-spec))))))
 530
 531 (defun char-db-insert-alist (alist &optional readable column)
 532   (unless column
 533     (setq column (current-column)))
 534   (let ((line-breaking
 535          (concat "\n" (make-string (1+ column) ?\ )))
 536         name value
 537         ret al ; cal
 538         key
 539         lbs cell rest separator)
 540     (insert "(")
 541     (while alist
 542       (setq name (car (car alist))
 543             value (cdr (car alist)))
 544       (cond ((eq name 'char)
 545              (insert "(char . ")
 546              (if (setq ret (condition-case nil
 547                                (find-char value)
 548                              (error nil)))
 549                  (progn
 550                    (setq al nil
 551                          ;; cal nil
 552                          )
 553                    (while value
 554                      (setq key (car (car value)))
 555                      ;; (if (find-charset key)
 556                      ;;     (setq cal (cons key cal))
 557                      (setq al (cons key al))
 558                      ;; )
 559                      (setq value (cdr value)))
 560                    (insert-char-attributes ret
 561                                            readable
 562                                            (or al 'none) ; cal
 563                                            nil 'for-sub-node))
 564                (insert (prin1-to-string value)))
 565              (insert ")")
 566              (insert line-breaking))
 567             ((consp value)
 568              (insert (format "(%-18s " name))
 569              (setq lbs (concat "\n" (make-string (current-column) ?\ )))
 570              (while (consp value)
 571                (setq cell (car value))
 572                (if (and (consp cell)
 573                         (consp (car cell))
 574                         (setq ret (condition-case nil
 575                                       (find-char cell)
 576                                     (error nil)))
 577                         )
 578                    (progn
 579                      (setq rest cell
 580                            al nil
 581                            ;; cal nil
 582                            )
 583                      (while rest
 584                        (setq key (car (car rest)))
 585                        ;; (if (find-charset key)
 586                        ;;     (setq cal (cons key cal))
 587                        (setq al (cons key al))
 588                        ;; )
 589                        (setq rest (cdr rest)))
 590                      (if separator
 591                          (insert lbs))
 592                      (insert-char-attributes ret
 593                                              readable
 594                                              al ; cal
 595                                              nil 'for-sub-node)
 596                      (setq separator lbs))
 597                  (if separator
 598                      (insert separator))
 599                  (insert (prin1-to-string cell))
 600                  (setq separator " "))
 601                (setq value (cdr value)))
 602              (insert ")")
 603              (insert line-breaking))
 604             (t
 605              (insert (format "(%-18s . %S)%s"
 606                              name value
 607                              line-breaking))))
 608       (setq alist (cdr alist))))
 609   (insert ")"))
 610
 611 (defun char-db-insert-char-reference (plist &optional readable column)
 612   (unless column
 613     (setq column (current-column)))
 614   (let ((line-breaking
 615          (concat "\n" (make-string (1+ column) ?\ )))
 616         (separator "")
 617         name value)
 618     (insert "(")
 619     (while plist
 620       (setq name (pop plist))
 621       (setq value (pop plist))
 622       (cond ((eq name :char)
 623              (insert separator)
 624              (insert ":char\t")
 625              (cond ((numberp value)
 626                     (setq value (decode-char '=ucs value)))
 627                    ;; ((consp value)
 628                    ;;  (setq value (or (find-char value)
 629                    ;;                  value)))
 630                    )
 631              (char-db-insert-char-spec value readable)
 632              (insert line-breaking)
 633              (setq separator ""))
 634             ((eq name :radical)
 635              (insert (format "%s%s\t%d ; %c%s"
 636                              separator
 637                              name value
 638                              (ideographic-radical value)
 639                              line-breaking))
 640              (setq separator ""))
 641             (t
 642              (insert (format "%s%s\t%S" separator name value))
 643              (setq separator line-breaking)))
 644       ))
 645   (insert ")"))
 646
 647 (defun char-db-decode-isolated-char (ccs code-point)
 648   (let (ret)
 649     (setq ret
 650           (cond ((eq ccs 'arabic-iso8859-6)
 651                  (decode-char ccs code-point))
 652                 ;; ((eq ccs '=gt)
 653                 ;;  (decode-builtin-char '==gt code-point))
 654                 ((and (memq ccs '(=gt-pj-1
 655                                   =gt-pj-2
 656                                   =gt-pj-3
 657                                   =gt-pj-4
 658                                   =gt-pj-5
 659                                   =gt-pj-6
 660                                   =gt-pj-7
 661                                   =gt-pj-8
 662                                   =gt-pj-9
 663                                   =gt-pj-10
 664                                   =gt-pj-11))
 665                       (setq ret (decode-char ccs code-point))
 666                       (setq ret (encode-char ret '=gt 'defined-only)))
 667                  (decode-builtin-char '=gt ret))
 668                 (t
 669                  (decode-builtin-char ccs code-point))))
 670     (cond ((null ret)
 671            (or (decode-char ccs code-point)
 672                (define-char (list (cons ccs code-point)))))
 673           ((and (<= 0 (char-int ret))
 674                 (<= (char-int ret) #x1F))
 675            (decode-char '=ucs (+ #x2400 (char-int ret))))
 676           ((= (char-int ret) #x7F)
 677            ?\u2421)
 678           (t ret))))
 679
 680 (defvar char-db-convert-obsolete-format t)
 681
 682 (defun char-db-insert-ccs-feature (name value line-breaking)
 683   (cond
 684    ((integerp value)
 685     (insert
 686      (format
 687       (cond
 688        ((memq name '(=>iwds-1
 689                      ==shinjigen
 690                      =shinjigen
 691                      =shinjigen@1ed ==shinjigen@1ed
 692                      =shinjigen@rev ==shinjigen@rev
 693                      =shinjigen/+p@rev ==shinjigen/+p@rev
 694                      ===daikanwa/ho ==daikanwa/ho
 695                      =daikanwa/ho =>>daikanwa/ho =>daikanwa/ho))
 696         "(%-18s .  %04d)\t; %c")
 697        ((eq name '=shinjigen@1ed/24pr)
 698         "(%-18s . %04d)\t; %c")
 699        ((or
 700          (memq name
 701                '(===daikanwa
 702                  ==daikanwa =daikanwa =>>daikanwa =>daikanwa
 703                  =daikanwa@rev1 =daikanwa@rev2
 704                  =daikanwa/+p ==daikanwa/+p ===daikanwa/+p
 705                  =>>daikanwa/+p
 706                  =daikanwa/+2p =>>daikanwa/+2p
 707                  =gt ==gt ===gt
 708                  =>>gt =+>gt =>gt
 709                  =gt-k ==gt-k ===gt-k
 710                  =>>gt-k =>gt-k
 711                  =adobe-japan1-0 ==adobe-japan1-0 ===adobe-japan1-0
 712                  =adobe-japan1-1 ==adobe-japan1-1 ===adobe-japan1-1
 713                  =adobe-japan1-2 ==adobe-japan1-2 ===adobe-japan1-2
 714                  =adobe-japan1-3 ==adobe-japan1-3 ===adobe-japan1-3
 715                  =adobe-japan1-4 ==adobe-japan1-4 ===adobe-japan1-4
 716                  =adobe-japan1-5 ==adobe-japan1-5 ===adobe-japan1-5
 717                  =adobe-japan1-6 ==adobe-japan1-6 ===adobe-japan1-6
 718                  =>>adobe-japan1-0 =+>adobe-japan1-0
 719                  =>>adobe-japan1-1 =+>adobe-japan1-1
 720                  =>>adobe-japan1-2 =+>adobe-japan1-2
 721                  =>>adobe-japan1-3 =+>adobe-japan1-3
 722                  =>>adobe-japan1-4 =+>adobe-japan1-4
 723                  =>>adobe-japan1-5 =+>adobe-japan1-5
 724                  =>>adobe-japan1-6 =+>adobe-japan1-6
 725                  =>cbeta =cbeta =>>cbeta ==cbeta ===cbeta
 726                  =zinbun-oracle =>zinbun-oracle
 727                  ===hng-jou ===hng-keg ===hng-dng ===hng-mam
 728                  ===hng-drt ===hng-kgk ===hng-myz ===hng-kda
 729                  ===hng-khi ===hng-khm ===hng-hok ===hng-kyd ===hng-sok
 730                  ===hng-yhk ===hng-kak ===hng-kar ===hng-kae
 731                  ===hng-sys ===hng-tsu ===hng-tzj
 732                  ===hng-hos ===hng-nak ===hng-jhk
 733                  ===hng-hod ===hng-gok ===hng-ink ===hng-nto
 734                  ===hng-nkm ===hng-k24 ===hng-nkk
 735                  ===hng-kcc ===hng-kcj ===hng-kbk ===hng-sik
 736                  ===hng-skk ===hng-kyu ===hng-ksk ===hng-wan
 737                  ===hng-okd ===hng-wad ===hng-kmi
 738                  ===hng-zkd ===hng-doh ===hng-jyu
 739                  ===hng-tzs ===hng-kss ===hng-kyo
 740                  ===hng-smk))
 741          ;; (string-match "^=adobe-" (symbol-name name))
 742          )
 743         "(%-18s . %05d)\t; %c")
 744        ((memq name '(=hanyo-denshi/ks
 745                      ==hanyo-denshi/ks ===hanyo-denshi/ks
 746                      =>>hanyo-denshi/ks
 747                      =koseki ==koseki
 748                      =mj ==mj ===mj =>>mj =>mj
 749                      =zihai mojikyo))
 750         "(%-18s . %06d)\t; %c")
 751        ((memq name '(=hanyo-denshi/tk ==hanyo-denshi/tk))
 752         "(%-18s . %08d)\t; %c")
 753        ((>= (charset-dimension name) 2)
 754         "(%-18s . #x%04X)\t; %c")
 755        (t
 756         "(%-18s . #x%02X)\t; %c"))
 757       name
 758       (if (= (charset-iso-graphic-plane name) 1)
 759           (logior value
 760                   (cond ((= (charset-dimension name) 1)
 761                          #x80)
 762                         ((= (charset-dimension name) 2)
 763                          #x8080)
 764                         ((= (charset-dimension name) 3)
 765                          #x808080)
 766                         (t 0)))
 767         value)
 768       (char-db-decode-isolated-char name value)))
 769     (if (and (= (charset-chars name) 94)
 770              (= (charset-dimension name) 2))
 771         (insert (format " [%02d-%02d]"
 772                         (- (lsh value -8) 32)
 773                         (- (logand value 255) 32))))
 774     )
 775    (t
 776     (insert (format "(%-18s . %s)" name value))
 777     ))
 778   (insert line-breaking))
 779
 780 (defun char-db-insert-relation-feature (char name value line-breaking
 781                                              ccss readable)
 782   (insert (format "(%-18s%s " name line-breaking))
 783   (let ((lbs (concat "\n" (make-string (current-column) ?\ )))
 784         separator cell sources required-features
 785         ret)
 786     (while (consp value)
 787       (setq cell (car value))
 788       (if (integerp cell)
 789           (setq cell (decode-char '=ucs cell)))
 790       (cond
 791        ((eq name '->subsumptive)
 792         (when (or (not (some (lambda (atr)
 793                                (get-char-attribute cell atr))
 794                              char-db-ignored-attributes))
 795                   (some (lambda (ccs)
 796                           (encode-char cell ccs 'defined-only))
 797                         ccss))
 798           (if separator
 799               (insert lbs))
 800           (let ((char-db-ignored-attributes
 801                  (cons '<-subsumptive
 802                        char-db-ignored-attributes)))
 803             (insert-char-attributes cell readable nil nil 'for-sub-node))
 804           (setq separator lbs))
 805         )
 806        ((characterp cell)
 807         (setq sources
 808               (get-char-attribute
 809                char (intern (format "%s*sources" name))))
 810         (setq required-features nil)
 811         (dolist (source sources)
 812           (cond
 813            ((memq source '(JP
 814                            JP/Jouyou
 815                            shinjigen shinjigen@1ed shinjigen@rev))
 816             (setq required-features
 817                   (union required-features
 818                          '(=jis-x0208
 819                            =jis-x0208@1990
 820                            =jis-x0213-1@2000
 821                            =jis-x0213-1@2004
 822                            =jis-x0213-2
 823                            =jis-x0212
 824                            =jis-x0208@1983
 825                            =jis-x0208@1978
 826                            =shinjigen
 827                            =shinjigen@1ed
 828                            =shinjigen@rev
 829                            =shinjigen/+p@rev))))
 830            ((eq source 'CN)
 831             (setq required-features
 832                   (union required-features
 833                          '(=gb2312
 834                            =gb12345
 835                            =iso-ir165)))))
 836           (cond
 837            ((find-charset (setq ret (intern (format "=%s" source))))
 838             (setq required-features
 839                   (cons ret required-features)))
 840            (t (setq required-features
 841                     (cons source required-features)))))
 842         (cond ((string-match "@JP" (symbol-name name))
 843                (setq required-features
 844                      (union required-features
 845                             '(=jis-x0208
 846                               =jis-x0208@1990
 847                               =jis-x0213-1-2000
 848                               =jis-x0213-2-2000
 849                               =jis-x0212
 850                               =jis-x0208@1983
 851                               =jis-x0208@1978))))
 852               ((string-match "@CN" (symbol-name name))
 853                (setq required-features
 854                      (union required-features
 855                             '(=gb2312
 856                               =gb12345
 857                               =iso-ir165)))))
 858         (if separator
 859             (insert lbs))
 860         (if readable
 861             (insert (format "%S" cell))
 862           (char-db-insert-char-spec cell readable
 863                                     nil
 864                                     required-features))
 865         (setq separator lbs))
 866        ((consp cell)
 867         (if separator
 868             (insert lbs))
 869         (if (consp (car cell))
 870             (char-db-insert-char-spec cell readable)
 871           (char-db-insert-char-reference cell readable))
 872         (setq separator lbs))
 873        (t
 874         (if separator
 875             (insert separator))
 876         (insert (prin1-to-string cell))
 877         (setq separator " ")))
 878       (setq value (cdr value)))
 879     (insert ")")
 880     (insert line-breaking)))
 881
 882 (defun insert-char-attributes (char &optional readable attributes column
 883                                     for-sub-node)
 884   (unless column
 885     (setq column (current-column)))
 886   (let (name value ; has-long-ccs-name
 887         rest
 888         radical strokes
 889         (line-breaking
 890          (concat "\n" (make-string (1+ column) ?\ )))
 891         lbs cell separator ret
 892         key al cal
 893         dest-ccss ; sources required-features
 894         ccss)
 895     (let (atr-d)
 896       (setq attributes
 897             (sort (if attributes
 898                       (if (consp attributes)
 899                           (progn
 900                             (dolist (name attributes)
 901                               (unless (memq name char-db-ignored-attributes)
 902                                 (if (find-charset name)
 903                                     (push name ccss))
 904                                 (push name atr-d)))
 905                             atr-d))
 906                     (dolist (name (char-attribute-list))
 907                       (unless (memq name char-db-ignored-attributes)
 908                         (if (find-charset name)
 909                             (push name ccss))
 910                         (push name atr-d)))
 911                     atr-d)
 912                   #'char-attribute-name<)))
 913     (insert "(")
 914     (when (memq '<-subsumptive attributes)
 915       (when (or readable (not for-sub-node))
 916         (when (setq value (get-char-attribute char '<-subsumptive))
 917           (char-db-insert-relation-feature char '<-subsumptive value
 918                                            line-breaking
 919                                            ccss readable)))
 920       (setq attributes (delq '<-subsumptive attributes)))
 921     (when (and (memq '<-denotational attributes)
 922                (setq value (get-char-attribute char '<-denotational)))
 923       (char-db-insert-relation-feature char '<-denotational value
 924                                        line-breaking
 925                                        ccss readable)
 926       (setq attributes (delq '<-denotational attributes)))
 927     (when (and (memq '<-denotational@component attributes)
 928                (setq value (get-char-attribute char '<-denotational@component)))
 929       (char-db-insert-relation-feature char '<-denotational@component value
 930                                        line-breaking
 931                                        ccss readable)
 932       (setq attributes (delq '<-denotational@component attributes)))
 933     (when (and (memq 'name attributes)
 934                (setq value (get-char-attribute char 'name)))
 935       (insert (format
 936                (if (> (+ (current-column) (length value)) 48)
 937                    "(name . %S)%s"
 938                  "(name               . %S)%s")
 939                value line-breaking))
 940       (setq attributes (delq 'name attributes))
 941       )
 942     (when (and (memq 'name* attributes)
 943                (setq value (get-char-attribute char 'name*)))
 944       (insert (format
 945                (if (> (+ (current-column) (length value)) 48)
 946                    "(name* . %S)%s"
 947                  "(name*              . %S)%s")
 948                value line-breaking))
 949       (setq attributes (delq 'name* attributes))
 950       )
 951     (when (and (memq 'script attributes)
 952                (setq value (get-char-attribute char 'script)))
 953       (insert (format "(script\t\t%s)%s"
 954                       (mapconcat (function prin1-to-string)
 955                                  value " ")
 956                       line-breaking))
 957       (setq attributes (delq 'script attributes))
 958       )
 959     (dolist (name '(=>ucs =>ucs*))
 960       (when (and (memq name attributes)
 961                  (setq value (get-char-attribute char name)))
 962         (insert (format "(%-18s . #x%04X)\t; %c%s"
 963                         name value (decode-char '=ucs value)
 964                         line-breaking))
 965         (setq attributes (delq name attributes))))
 966     (dolist (name '(=>ucs@gb =>ucs@big5))
 967       (when (and (memq name attributes)
 968                  (setq value (get-char-attribute char name)))
 969         (insert (format "(%-18s . #x%04X)\t; %c%s"
 970                         name value
 971                         (decode-char (intern
 972                                       (concat "="
 973                                               (substring
 974                                                (symbol-name name) 2)))
 975                                      value)
 976                         line-breaking))
 977         (setq attributes (delq name attributes))
 978         ))
 979     ;; (dolist (name '(=>daikanwa))
 980     ;;   (when (and (memq name attributes)
 981     ;;              (setq value (get-char-attribute char name)))
 982     ;;     (insert
 983     ;;      (if (integerp value)
 984     ;;          (format "(%-18s . %05d)\t; %c%s"
 985     ;;                  name value (decode-char '=daikanwa value)
 986     ;;                  line-breaking)
 987     ;;        (format "(%-18s %s)\t; %c%s"
 988     ;;                name
 989     ;;                (mapconcat (function prin1-to-string)
 990     ;;                           value " ")
 991     ;;                (char-representative-of-daikanwa char)
 992     ;;                line-breaking)))
 993     ;;     (setq attributes (delq name attributes))))
 994     (when (and (memq 'general-category attributes)
 995                (setq value (get-char-attribute char 'general-category)))
 996       (insert (format
 997                "(general-category\t%s) ; %s%s"
 998                (mapconcat (lambda (cell)
 999                             (format "%S" cell))
1000                           value " ")
1001                (cond ((rassoc value unidata-normative-category-alist)
1002                       "Normative Category")
1003                      ((rassoc value unidata-informative-category-alist)
1004                       "Informative Category")
1005                      (t
1006                       "Unknown Category"))
1007                line-breaking))
1008       (setq attributes (delq 'general-category attributes))
1009       )
1010     (when (and (memq 'bidi-category attributes)
1011                (setq value (get-char-attribute char 'bidi-category)))
1012       (insert (format "(bidi-category\t. %S)%s"
1013                       value
1014                       line-breaking))
1015       (setq attributes (delq 'bidi-category attributes))
1016       )
1017     (unless (or (not (memq 'mirrored attributes))
1018                 (eq (setq value (get-char-attribute char 'mirrored 'empty))
1019                     'empty))
1020       (insert (format "(mirrored\t\t. %S)%s"
1021                       value
1022                       line-breaking))
1023       (setq attributes (delq 'mirrored attributes))
1024       )
1025     (cond
1026      ((and (memq 'decimal-digit-value attributes)
1027            (setq value (get-char-attribute char 'decimal-digit-value)))
1028       (insert (format "(decimal-digit-value . %S)%s"
1029                       value
1030                       line-breaking))
1031       (setq attributes (delq 'decimal-digit-value attributes))
1032       (when (and (memq 'digit-value attributes)
1033                  (setq value (get-char-attribute char 'digit-value)))
1034         (insert (format "(digit-value\t . %S)%s"
1035                         value
1036                         line-breaking))
1037         (setq attributes (delq 'digit-value attributes))
1038         )
1039       (when (and (memq 'numeric-value attributes)
1040                  (setq value (get-char-attribute char 'numeric-value)))
1041         (insert (format "(numeric-value\t . %S)%s"
1042                         value
1043                         line-breaking))
1044         (setq attributes (delq 'numeric-value attributes))
1045         )
1046       )
1047      (t
1048       (when (and (memq 'digit-value attributes)
1049                  (setq value (get-char-attribute char 'digit-value)))
1050         (insert (format "(digit-value\t. %S)%s"
1051                         value
1052                         line-breaking))
1053         (setq attributes (delq 'digit-value attributes))
1054         )
1055       (when (and (memq 'numeric-value attributes)
1056                  (setq value (get-char-attribute char 'numeric-value)))
1057         (insert (format "(numeric-value\t. %S)%s"
1058                         value
1059                         line-breaking))
1060         (setq attributes (delq 'numeric-value attributes))
1061         )))
1062     (when (and (memq 'iso-10646-comment attributes)
1063                (setq value (get-char-attribute char 'iso-10646-comment)))
1064       (insert (format "(iso-10646-comment\t. %S)%s"
1065                       value
1066                       line-breaking))
1067       (setq attributes (delq 'iso-10646-comment attributes))
1068       )
1069     (when (and (memq 'morohashi-daikanwa attributes)
1070                (setq value (get-char-attribute char 'morohashi-daikanwa)))
1071       (insert (format "(morohashi-daikanwa\t%s)%s"
1072                       (mapconcat (function prin1-to-string) value " ")
1073                       line-breaking))
1074       (setq attributes (delq 'morohashi-daikanwa attributes))
1075       )
1076     (setq radical nil
1077           strokes nil)
1078     (when (and (memq 'ideographic-radical attributes)
1079                (setq value (get-char-attribute char 'ideographic-radical)))
1080       (setq radical value)
1081       (insert (format "(ideographic-radical . %S)\t; %c%s"
1082                       radical
1083                       (ideographic-radical radical)
1084                       line-breaking))
1085       (setq attributes (delq 'ideographic-radical attributes))
1086       )
1087     (when (and (memq 'shuowen-radical attributes)
1088                (setq value (get-char-attribute char 'shuowen-radical)))
1089       (insert (format "(shuowen-radical\t. %S)\t; %c%s"
1090                       value
1091                       (shuowen-radical value)
1092                       line-breaking))
1093       (setq attributes (delq 'shuowen-radical attributes))
1094       )
1095     (let (key)
1096       (dolist (domain
1097                (append
1098                 char-db-feature-domains
1099                 (let (dest domain)
1100                   (dolist (feature (char-attribute-list))
1101                     (setq feature (symbol-name feature))
1102                     (when (string-match
1103                            "\\(radical\\|strokes\\)@\\([^@*]+\\)\\(\\*\\|$\\)"
1104                            feature)
1105                       (setq domain (intern (match-string 2 feature)))
1106                      (unless (memq domain dest)
1107                        (setq dest (cons domain dest)))))
1108                   (sort dest #'string<))))
1109         (setq key (intern (format "%s@%s" 'ideographic-radical domain)))
1110         (when (and (memq key attributes)
1111                    (setq value (get-char-attribute char key)))
1112           (setq radical value)
1113           (insert (format "(%s . %S)\t; %c%s"
1114                           key
1115                           radical
1116                           (ideographic-radical radical)
1117                           line-breaking))
1118           (setq attributes (delq key attributes))
1119           )
1120         (setq key (intern (format "%s@%s" 'ideographic-strokes domain)))
1121         (when (and (memq key attributes)
1122                    (setq value (get-char-attribute char key)))
1123           (setq strokes value)
1124           (insert (format "(%s . %S)%s"
1125                           key
1126                           strokes
1127                           line-breaking))
1128           (setq attributes (delq key attributes))
1129           )
1130         (setq key (intern (format "%s@%s" 'total-strokes domain)))
1131         (when (and (memq key attributes)
1132                    (setq value (get-char-attribute char key)))
1133           (insert (format "(%s       . %S)%s"
1134                           key
1135                           value
1136                           line-breaking))
1137           (setq attributes (delq key attributes))
1138           )
1139         (dolist (feature '(ideographic-radical
1140                            ideographic-strokes
1141                            total-strokes))
1142           (setq key (intern (format "%s@%s*sources" feature domain)))
1143           (when (and (memq key attributes)
1144                      (setq value (get-char-attribute char key)))
1145             (insert (format "(%s%s" key line-breaking))
1146             (dolist (cell value)
1147               (insert (format " %s" cell)))
1148             (insert ")")
1149             (insert line-breaking)
1150             (setq attributes (delq key attributes))
1151             ))
1152         ))
1153     (when (and (memq 'ideographic-strokes attributes)
1154                (setq value (get-char-attribute char 'ideographic-strokes)))
1155       (setq strokes value)
1156       (insert (format "(ideographic-strokes . %S)%s"
1157                       strokes
1158                       line-breaking))
1159       (setq attributes (delq 'ideographic-strokes attributes))
1160       )
1161     (when (and (memq 'kangxi-radical attributes)
1162                (setq value (get-char-attribute char 'kangxi-radical)))
1163       (unless (eq value radical)
1164         (insert (format "(kangxi-radical\t . %S)\t; %c%s"
1165                         value
1166                         (ideographic-radical value)
1167                         line-breaking))
1168         (or radical
1169             (setq radical value)))
1170       (setq attributes (delq 'kangxi-radical attributes))
1171       )
1172     (when (and (memq 'kangxi-strokes attributes)
1173                (setq value (get-char-attribute char 'kangxi-strokes)))
1174       (unless (eq value strokes)
1175         (insert (format "(kangxi-strokes\t . %S)%s"
1176                         value
1177                         line-breaking))
1178         (or strokes
1179             (setq strokes value)))
1180       (setq attributes (delq 'kangxi-strokes attributes))
1181       )
1182     ;; (when (and (memq 'japanese-radical attributes)
1183     ;;            (setq value (get-char-attribute char 'japanese-radical)))
1184     ;;   (unless (eq value radical)
1185     ;;     (insert (format "(japanese-radical\t . %S)\t; %c%s"
1186     ;;                     value
1187     ;;                     (ideographic-radical value)
1188     ;;                     line-breaking))
1189     ;;     (or radical
1190     ;;         (setq radical value)))
1191     ;;   (setq attributes (delq 'japanese-radical attributes))
1192     ;;   )
1193     (when (and (memq 'japanese-strokes attributes)
1194                (setq value (get-char-attribute char 'japanese-strokes)))
1195       (unless (eq value strokes)
1196         (insert (format "(japanese-strokes\t . %S)%s"
1197                         value
1198                         line-breaking))
1199         (or strokes
1200             (setq strokes value)))
1201       (setq attributes (delq 'japanese-strokes attributes))
1202       )
1203     (when (and (memq 'cns-radical attributes)
1204                (setq value (get-char-attribute char 'cns-radical)))
1205       (insert (format "(cns-radical\t . %S)\t; %c%s"
1206                       value
1207                       (ideographic-radical value)
1208                       line-breaking))
1209       (setq attributes (delq 'cns-radical attributes))
1210       )
1211     (when (and (memq 'cns-strokes attributes)
1212                (setq value (get-char-attribute char 'cns-strokes)))
1213       (unless (eq value strokes)
1214         (insert (format "(cns-strokes\t . %S)%s"
1215                         value
1216                         line-breaking))
1217         (or strokes
1218             (setq strokes value)))
1219       (setq attributes (delq 'cns-strokes attributes))
1220       )
1221     ;; (when (and (memq 'shinjigen-1-radical attributes)
1222     ;;            (setq value (get-char-attribute char 'shinjigen-1-radical)))
1223     ;;   (unless (eq value radical)
1224     ;;     (insert (format "(shinjigen-1-radical . %S)\t; %c%s"
1225     ;;                     value
1226     ;;                     (ideographic-radical value)
1227     ;;                     line-breaking))
1228     ;;     (or radical
1229     ;;         (setq radical value)))
1230     ;;   (setq attributes (delq 'shinjigen-1-radical attributes))
1231     ;;   )
1232     ;; (when (and (memq 'ideographic- attributes)
1233     ;;            (setq value (get-char-attribute char 'ideographic-)))
1234     ;;   (insert "(ideographic-       ")
1235     ;;   (setq lbs (concat "\n" (make-string (current-column) ?\ ))
1236     ;;         separator nil)
1237     ;;   (while (consp value)
1238     ;;     (setq cell (car value))
1239     ;;     (if (integerp cell)
1240     ;;         (setq cell (decode-char '=ucs cell)))
1241     ;;     (cond ((characterp cell)
1242     ;;            (if separator
1243     ;;                (insert lbs))
1244     ;;            (if readable
1245     ;;                (insert (format "%S" cell))
1246     ;;              (char-db-insert-char-spec cell readable))
1247     ;;            (setq separator lbs))
1248     ;;           ((consp cell)
1249     ;;            (if separator
1250     ;;                (insert lbs))
1251     ;;            (if (consp (car cell))
1252     ;;                (char-db-insert-char-spec cell readable)
1253     ;;              (char-db-insert-char-reference cell readable))
1254     ;;            (setq separator lbs))
1255     ;;           (t
1256     ;;            (if separator
1257     ;;                (insert separator))
1258     ;;            (insert (prin1-to-string cell))
1259     ;;            (setq separator " ")))
1260     ;;     (setq value (cdr value)))
1261     ;;   (insert ")")
1262     ;;   (insert line-breaking)
1263     ;;   (setq attributes (delq 'ideographic- attributes)))
1264     (when (and (memq 'total-strokes attributes)
1265                (setq value (get-char-attribute char 'total-strokes)))
1266       (insert (format "(total-strokes       . %S)%s"
1267                       value
1268                       line-breaking))
1269       (setq attributes (delq 'total-strokes attributes))
1270       )
1271     ;; (when (and (memq '->ideograph attributes)
1272     ;;            (setq value (get-char-attribute char '->ideograph)))
1273     ;;   (insert (format "(->ideograph\t%s)%s"
1274     ;;                   (mapconcat (lambda (code)
1275     ;;                                (cond ((symbolp code)
1276     ;;                                       (symbol-name code))
1277     ;;                                      ((integerp code)
1278     ;;                                       (format "#x%04X" code))
1279     ;;                                      (t
1280     ;;                                       (format "%s %S"
1281     ;;                                               line-breaking code))))
1282     ;;                              value " ")
1283     ;;                   line-breaking))
1284     ;;   (setq attributes (delq '->ideograph attributes))
1285     ;;   )
1286     ;; (when (and (memq '->decomposition attributes)
1287     ;;            (setq value (get-char-attribute char '->decomposition)))
1288     ;;   (insert (format "(->decomposition\t%s)%s"
1289     ;;                   (mapconcat (lambda (code)
1290     ;;                                (cond ((symbolp code)
1291     ;;                                       (symbol-name code))
1292     ;;                                      ((characterp code)
1293     ;;                                       (if readable
1294     ;;                                           (format "%S" code)
1295     ;;                                         (format "#x%04X"
1296     ;;                                                 (char-int code))
1297     ;;                                         ))
1298     ;;                                      ((integerp code)
1299     ;;                                       (format "#x%04X" code))
1300     ;;                                      (t
1301     ;;                                       (format "%s%S" line-breaking code))))
1302     ;;                              value " ")
1303     ;;                   line-breaking))
1304     ;;   (setq attributes (delq '->decomposition attributes))
1305     ;;   )
1306     (if (equal (get-char-attribute char '->titlecase)
1307                (get-char-attribute char '->uppercase))
1308         (setq attributes (delq '->titlecase attributes)))
1309     (when (and (memq '->mojikyo attributes)
1310                (setq value (get-char-attribute char '->mojikyo)))
1311       (insert (format "(->mojikyo\t\t. %06d)\t; %c%s"
1312                       value (decode-char 'mojikyo value)
1313                       line-breaking))
1314       (setq attributes (delq '->mojikyo attributes))
1315       )
1316     ;; (when (and (memq 'hanyu-dazidian-vol attributes)
1317     ;;            (setq value (get-char-attribute char 'hanyu-dazidian-vol)))
1318     ;;   (insert (format "(hanyu-dazidian-vol  . %d)%s"
1319     ;;                   value line-breaking))
1320     ;;   (setq attributes (delq 'hanyu-dazidian-vol attributes))
1321     ;;   )
1322     ;; (when (and (memq 'hanyu-dazidian-page attributes)
1323     ;;            (setq value (get-char-attribute char 'hanyu-dazidian-page)))
1324     ;;   (insert (format "(hanyu-dazidian-page . %d)%s"
1325     ;;                   value line-breaking))
1326     ;;   (setq attributes (delq 'hanyu-dazidian-page attributes))
1327     ;;   )
1328     ;; (when (and (memq 'hanyu-dazidian-char attributes)
1329     ;;            (setq value (get-char-attribute char 'hanyu-dazidian-char)))
1330     ;;   (insert (format "(hanyu-dazidian-char . %d)%s"
1331     ;;                   value line-breaking))
1332     ;;   (setq attributes (delq 'hanyu-dazidian-char attributes))
1333     ;;   )
1334     (unless readable
1335       (dolist (ignored '(composition
1336                          ->denotational <-subsumptive ->ucs-unified
1337                          ->ideographic-component-forms))
1338         (setq attributes (delq ignored attributes))))
1339     (while attributes
1340       (setq name (car attributes))
1341       (unless (eq (setq value (get-char-attribute char name 'value-is-empty))
1342                   'value-is-empty)
1343         (cond ((setq ret (find-charset name))
1344                (setq name (charset-name ret))
1345                (when (not (memq name dest-ccss))
1346                  (setq dest-ccss (cons name dest-ccss))
1347                  (char-db-insert-ccs-feature name value line-breaking))
1348                )
1349               ((string-match "^=>ucs@" (symbol-name name))
1350                (insert (format "(%-18s . #x%04X)\t; %c%s"
1351                                name value (decode-char '=ucs value)
1352                                line-breaking))
1353                )
1354               ((eq name 'jisx0208-1978/4X)
1355                (insert (format "(%-18s . #x%04X)%s"
1356                                name value
1357                                line-breaking))
1358                )
1359               ((and
1360                 (not readable)
1361                 (not (eq name '->subsumptive))
1362                 (not (eq name '->uppercase))
1363                 (not (eq name '->lowercase))
1364                 (not (eq name '->titlecase))
1365                 (not (eq name '->canonical))
1366                 (not (eq name '->Bopomofo))
1367                 (not (eq name '->mistakable))
1368                 (not (eq name '->ideographic-variants))
1369                 (null (get-char-attribute
1370                        char (intern (format "%s*sources" name))))
1371                 (not (string-match "\\*sources$" (symbol-name name)))
1372                 (null (get-char-attribute
1373                        char (intern (format "%s*note" name))))
1374                 (not (string-match "\\*note$" (symbol-name name)))
1375                 (or (eq name '<-identical)
1376                     (eq name '<-uppercase)
1377                     (eq name '<-lowercase)
1378                     (eq name '<-titlecase)
1379                     (eq name '<-canonical)
1380                     (eq name '<-ideographic-variants)
1381                     ;; (eq name '<-synonyms)
1382                     (string-match "^<-synonyms" (symbol-name name))
1383                     (eq name '<-mistakable)
1384                     (when (string-match "^->" (symbol-name name))
1385                       (cond
1386                        ((string-match "^->fullwidth" (symbol-name name))
1387                         (not (and (consp value)
1388                                   (characterp (car value))
1389                                   (encode-char
1390                                    (car value) '=ucs 'defined-only)))
1391                         )
1392                        (t)))
1393                     ))
1394                )
1395               ((or (eq name 'ideographic-structure)
1396                    (eq name 'ideographic-combination)
1397                    ;; (eq name 'ideographic-)
1398                    (eq name '=decomposition)
1399                    (char-feature-base-name= '=decomposition name)
1400                    (char-feature-base-name= '=>decomposition name)
1401                    ;; (string-match "^=>*decomposition\\(@[^*]+\\)?$"
1402                    ;;               (symbol-name name))
1403                    (string-match "^\\(->\\|<-\\)[^*]*$" (symbol-name name))
1404                    (string-match "^\\(->\\|<-\\)[^*]*\\*sources$"
1405                                  (symbol-name name))
1406                    )
1407                (char-db-insert-relation-feature char name value
1408                                                 line-breaking
1409                                                 ccss readable))
1410               ;; ((memq name '(ideograph=
1411               ;;               original-ideograph-of
1412               ;;               ancient-ideograph-of
1413               ;;               vulgar-ideograph-of
1414               ;;               wrong-ideograph-of
1415               ;;               ;; simplified-ideograph-of
1416               ;;               ideographic-variants
1417               ;;               ;; ideographic-different-form-of
1418               ;;               ))
1419               ;;  (insert (format "(%-18s%s " name line-breaking))
1420               ;;  (setq lbs (concat "\n" (make-string (current-column) ?\ ))
1421               ;;        separator nil)
1422               ;;  (while (consp value)
1423               ;;    (setq cell (car value))
1424               ;;    (if (and (consp cell)
1425               ;;             (consp (car cell)))
1426               ;;        (progn
1427               ;;          (if separator
1428               ;;              (insert lbs))
1429               ;;          (char-db-insert-alist cell readable)
1430               ;;          (setq separator lbs))
1431               ;;      (if separator
1432               ;;          (insert separator))
1433               ;;      (insert (prin1-to-string cell))
1434               ;;      (setq separator " "))
1435               ;;    (setq value (cdr value)))
1436               ;;  (insert ")")
1437               ;;  (insert line-breaking))
1438               ((consp value)
1439                (insert (format "(%-18s " name))
1440                (setq lbs (concat "\n" (make-string (current-column) ?\ ))
1441                      separator nil)
1442                (while (consp value)
1443                  (setq cell (car value))
1444                  (if (and (consp cell)
1445                           (consp (car cell))
1446                           (setq ret (condition-case nil
1447                                         (find-char cell)
1448                                       (error nil))))
1449                      (progn
1450                        (setq rest cell
1451                              al nil
1452                              cal nil)
1453                        (while rest
1454                          (setq key (car (car rest)))
1455                          (if (find-charset key)
1456                              (setq cal (cons key cal))
1457                            (setq al (cons key al)))
1458                          (setq rest (cdr rest)))
1459                        (if separator
1460                            (insert lbs))
1461                        (insert-char-attributes ret
1462                                                readable
1463                                                al ; cal
1464                                                nil 'for-sub-node)
1465                        (setq separator lbs))
1466                    (setq ret (prin1-to-string cell))
1467                    (if separator
1468                        (if (< (+ (current-column)
1469                                  (length ret)
1470                                  (length separator))
1471                               76)
1472                            (insert separator)
1473                          (insert lbs)))
1474                    (insert ret)
1475                    (setq separator " "))
1476                  (setq value (cdr value)))
1477                (insert ")")
1478                (insert line-breaking))
1479               (t
1480                (insert (format "(%-18s" name))
1481                (setq ret (prin1-to-string value))
1482                (unless (< (+ (current-column)
1483                              (length ret)
1484                              3)
1485                           76)
1486                  (insert line-breaking))
1487                (insert " . " ret ")" line-breaking)
1488                ;; (insert (format "(%-18s . %S)%s"
1489                ;;                 name value
1490                ;;                 line-breaking))
1491                )
1492               ))
1493       (setq attributes (cdr attributes)))
1494     (insert ")")))
1495
1496 (defun insert-char-data (char &optional readable
1497                               attributes)
1498   (save-restriction
1499     (narrow-to-region (point)(point))
1500     (insert "(define-char
1501   '")
1502     (insert-char-attributes char readable attributes)
1503     (insert ")\n")
1504     (goto-char (point-min))
1505     (while (re-search-forward "[ \t]+$" nil t)
1506       (replace-match ""))
1507     ;; from tabify.
1508     (goto-char (point-min))
1509     (while (re-search-forward "[ \t][ \t][ \t]*" nil t)
1510       (let ((column (current-column))
1511             (indent-tabs-mode t))
1512         (delete-region (match-beginning 0) (point))
1513         (indent-to column)))
1514     (goto-char (point-max))
1515     ;; (tabify (point-min)(point-max))
1516     ))
1517
1518 (defun insert-char-data-with-variant (char &optional printable
1519                                            no-ucs-unified
1520                                            script excluded-script)
1521   (insert-char-data char printable)
1522   (let ((variants (char-variants char))
1523         rest
1524         variant vs ret)
1525     (setq variants (sort variants #'<))
1526     (setq rest variants)
1527     (setq variants (cons char variants))
1528     (while rest
1529       (setq variant (car rest))
1530       (unless (get-char-attribute variant '<-subsumptive)
1531         (if (and (or (null script)
1532                      (null (setq vs (get-char-attribute variant 'script)))
1533                      (memq script vs))
1534                  (or (null excluded-script)
1535                      (null (setq vs (get-char-attribute variant 'script)))
1536                      (not (memq excluded-script vs))))
1537             (unless (and no-ucs-unified (get-char-attribute variant '=ucs))
1538               (insert-char-data variant printable)
1539               (if (setq ret (char-variants variant))
1540                   (while ret
1541                     (or (memq (car ret) variants)
1542                         ;; (get-char-attribute (car ret) '<-subsumptive)
1543                         (setq rest (nconc rest (list (car ret)))))
1544                     (setq ret (cdr ret)))))))
1545       (setq rest (cdr rest)))))
1546
1547 (defun insert-char-range-data (min max &optional script excluded-script)
1548   (let ((code min)
1549         char)
1550     (while (<= code max)
1551       (setq char (decode-char '=ucs code))
1552       (if (encode-char char '=ucs 'defined-only)
1553           (insert-char-data-with-variant char nil 'no-ucs-unified
1554                                          script excluded-script))
1555       (setq code (1+ code)))))
1556
1557 (defun write-char-range-data-to-file (min max file
1558                                           &optional script excluded-script)
1559   (let ((coding-system-for-write char-db-file-coding-system))
1560     (with-temp-buffer
1561       (insert (format ";; -*- coding: %s -*-\n"
1562                       char-db-file-coding-system))
1563       (insert-char-range-data min max script excluded-script)
1564       (write-region (point-min)(point-max) file))))
1565
1566 (defvar what-character-original-window-configuration)
1567
1568 ;;;###autoload
1569 (defun what-char-definition (char)
1570   (interactive (list (char-after)))
1571   (let ((buf (get-buffer-create "*Character Description*"))
1572         (the-buf (current-buffer))
1573         (win-conf (current-window-configuration)))
1574     (pop-to-buffer buf)
1575     (make-local-variable 'what-character-original-window-configuration)
1576     (setq what-character-original-window-configuration win-conf)
1577     (setq buffer-read-only nil)
1578     (erase-buffer)
1579     (condition-case err
1580         (progn
1581           (insert-char-data-with-variant char 'printable)
1582           (unless (char-attribute-alist char)
1583             (insert (format ";; = %c\n"
1584                             (let* ((rest (split-char char))
1585                                    (ccs (pop rest))
1586                                    (code (pop rest)))
1587                               (while rest
1588                                 (setq code (logior (lsh code 8)
1589                                                    (pop rest))))
1590                               (decode-char ccs code)))))
1591           ;; (char-db-update-comment)
1592           (set-buffer-modified-p nil)
1593           (view-mode the-buf (lambda (buf)
1594                                (set-window-configuration
1595                                 what-character-original-window-configuration)
1596                                ))
1597           (goto-char (point-min)))
1598       (error (progn
1599                (set-window-configuration
1600                 what-character-original-window-configuration)
1601                (signal (car err) (cdr err)))))))
1602
1603
1604 ;;; @ end
1605 ;;;
1606
1607 (provide 'char-db-util)
1608
1609 ;;; char-db-util.el ends here