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 ;;   2008,2009,2010,2011,2012,2013,2014,2015,2016,2017,2018,2019 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 ?𧮫 ?只 ?㕯 ; 050
  72    ?句 ?丩 ?古 ?十 ?卅 ?言 ?誩 ?音 ?䇂 ?丵 ; 060
  73    ?菐 ?廾 ?𠬜 ?共 ?異 ?舁 ?𦥑 ?䢅 ?爨 ?革 ; 070
  74    ?鬲 ?䰜 ?爪 ?𠃨 ?鬥 ?又 ?𠂇 ?史 ?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    ?雈 ?𦫳 ?𥄕 ?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 ?丯 ; 140
  81    ?耒 ?角 ?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 ?𠧪 ?齊 ?朿 ?片 ?鼎 ; 250
  93    ?克 ?彔 ?禾 ?秝 ?黍 ?香 ?米 ?毇 ?臼 ?凶 ; 260
  94    ?𣎳 ?𣏟 ?麻 ?尗 ?耑 ?韭 ?瓜 ?瓠 ?宀 ?宮 ; 270
  95    ?呂 ?穴 ?㝱 ?疒 ?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    ?夨 ?夭 ?交 ?尣 ?壺 ?壹 ?㚔 ?奢 ?亢 ?夲 ; 400
 108    ])
 109
 110 (defun shuowen-radical (number)
 111   (aref shuowen-radicals (1- number)))
 112
 113 (defvar char-db-file-coding-system 'utf-8-mcs-er)
 114
 115 (defvar char-db-ignored-attributes '(ideographic-products))
 116
 117 (defvar char-db-coded-charset-priority-list
 118   '(ascii
 119     control-1
 120     latin-iso8859-1
 121     latin-iso8859-2
 122     latin-iso8859-3
 123     latin-iso8859-4
 124     latin-iso8859-9
 125     latin-jisx0201
 126     cyrillic-iso8859-5
 127     greek-iso8859-7
 128     thai-tis620
 129     ;; =mj
 130     ;; =adobe-japan1-0
 131     ;; =adobe-japan1-1
 132     ;; =adobe-japan1-2
 133     ;; =adobe-japan1-3
 134     ;; =adobe-japan1-4
 135     ;; =adobe-japan1-5
 136     ;; =adobe-japan1-6
 137     =jis-x0208
 138     =jis-x0208@1978
 139     =jis-x0208@1983
 140     japanese-jisx0212
 141     chinese-gb2312
 142     =jis-x0208@1990
 143     chinese-cns11643-1
 144     chinese-cns11643-2
 145     chinese-cns11643-3
 146     chinese-cns11643-4
 147     chinese-cns11643-5
 148     chinese-cns11643-6
 149     chinese-cns11643-7
 150     =jis-x0213-1
 151     =jis-x0213-1@2000
 152     =jis-x0213-1@2004
 153     =jis-x0213-2
 154     korean-ksc5601
 155     chinese-isoir165
 156     katakana-jisx0201
 157     hebrew-iso8859-8
 158     chinese-gb12345
 159     latin-viscii
 160     ethiopic-ucs
 161     =big5-cdp
 162     =hanyo-denshi/ja
 163     =hanyo-denshi/jb
 164     =hanyo-denshi/jc
 165     =hanyo-denshi/jd
 166     =hanyo-denshi/ft
 167     =hanyo-denshi/ia
 168     =hanyo-denshi/ib
 169     =hanyo-denshi/hg
 170     =hanyo-denshi/jt
 171     =hanyo-denshi/ks
 172     =hanyo-denshi/tk
 173     ;; ==mj
 174     ;; ==adobe-japan1-0
 175     ;; ==adobe-japan1-1
 176     ;; ==adobe-japan1-2
 177     ;; ==adobe-japan1-3
 178     ;; ==adobe-japan1-4
 179     ;; ==adobe-japan1-5
 180     ;; ==adobe-japan1-6
 181     ==jis-x0208
 182     ==jis-x0213-1
 183     ==jis-x0213-2
 184     ==jis-x0212
 185     ==hanyo-denshi/ja
 186     ==hanyo-denshi/jb
 187     ==hanyo-denshi/jc
 188     ==hanyo-denshi/ft
 189     ==hanyo-denshi/ia
 190     ==hanyo-denshi/ib
 191     ==hanyo-denshi/hg
 192     ==hanyo-denshi/jt
 193     ==hanyo-denshi/ks
 194     ==hanyo-denshi/tk
 195     =gt
 196     =gt-k
 197     =daikanwa
 198     =daikanwa@rev2
 199     =daikanwa@rev1
 200     =daikanwa/+p
 201     ==daikanwa
 202     =cbeta
 203     ideograph-hanziku-1
 204     ideograph-hanziku-2
 205     ideograph-hanziku-3
 206     ideograph-hanziku-4
 207     ideograph-hanziku-5
 208     ideograph-hanziku-6
 209     ideograph-hanziku-7
 210     ideograph-hanziku-8
 211     ideograph-hanziku-9
 212     ideograph-hanziku-10
 213     ideograph-hanziku-11
 214     ideograph-hanziku-12
 215     ;; =>>>adobe-japan1-0
 216     ;; =>>>adobe-japan1-1
 217     ;; =>>>adobe-japan1-2
 218     ;; =>>>adobe-japan1-3
 219     ;; =>>>adobe-japan1-4
 220     ;; =>>>adobe-japan1-5
 221     ;; =>>>adobe-japan1-6
 222     ;; =>>>jis-x0208
 223     ;; =>>>jis-x0213-1
 224     ;; =>>>jis-x0213-2
 225     ;; =>>>hanyo-denshi/ja
 226     ;; =>>>hanyo-denshi/jb
 227     ;; =>>>hanyo-denshi/jc
 228     ;; =>>>hanyo-denshi/ft
 229     ;; =>>>hanyo-denshi/ib
 230     ;; =>>>hanyo-denshi/hg
 231     ;; =>>>hanyo-denshi/jt
 232     ;; =>>>hanyo-denshi/ks
 233     ;; =>>>gt
 234     =>>adobe-japan1-0
 235     =>>adobe-japan1-1
 236     =>>adobe-japan1-2
 237     =>>adobe-japan1-3
 238     =>>adobe-japan1-4
 239     =>>adobe-japan1-5
 240     =>>adobe-japan1-6
 241     =>>jis-x0208
 242     =>>jis-x0213-1
 243     =>>jis-x0213-1@2000
 244     =>>jis-x0213-1@2004
 245     =>>jis-x0213-2
 246     =>>jis-x0208@1978
 247     =>>hanyo-denshi/ft
 248     =>>hanyo-denshi/jt
 249     =>>hanyo-denshi/ks
 250     =>>gt
 251     =>>daikanwa
 252     =>>cbeta
 253     =+>jis-x0208
 254     =+>jis-x0213-1
 255     =+>jis-x0213-2
 256     =+>adobe-japan1-0
 257     =+>adobe-japan1-1
 258     =+>adobe-japan1-2
 259     =+>adobe-japan1-3
 260     =+>adobe-japan1-4
 261     =+>adobe-japan1-5
 262     =+>adobe-japan1-6
 263     =+>jis-x0208@1978
 264     =>jis-x0208
 265     =>jis-x0208@1997
 266     =>jis-x0213-1
 267     =>jis-x0213-1@2000
 268     =>jis-x0213-1@2004
 269     =>jis-x0213-2
 270     ==>ucs@bucs
 271     ==>daijiten
 272     =>iwds-1
 273     ;; =>ucs@hanyo-denshi
 274     =>ucs@iso
 275     =>ucs@unicode
 276     =>ucs@jis
 277     =>ucs@cns
 278     =>ucs@ks
 279     =+>ucs@iso
 280     =+>ucs@unicode
 281     =+>ucs@jis
 282     =+>ucs@jis/1990
 283     =+>ucs@cns
 284     =+>ucs@ks
 285     =>>ucs@iso
 286     =>>ucs@unicode
 287     =>>ucs@jis
 288     =>>ucs@cns
 289     =>>>ucs@iso
 290     =>>>ucs@unicode
 291     ==ucs@iso
 292     ==ucs@unicode
 293     ;; ==ucs@cns
 294     ==gb2312
 295     ==ks-x1001
 296     ==cns11643-1
 297     ==cns11643-2
 298     ==cns11643-3
 299     ==cns11643-4
 300     ==cns11643-5
 301     ==cns11643-6
 302     ==cns11643-7
 303     ==gt
 304     ==jis-x0208@1990
 305     ;; ==jis-x0208@1983
 306     ==jis-x0208@1978
 307     ==gt-k
 308     =ucs@iso
 309     =ucs@unicode
 310     =ucs@cns
 311     ==big5-cdp
 312     ==cbeta
 313     =>>big5-cdp
 314     =>>gt-k
 315     =+>gt
 316     =+>big5-cdp
 317     =>gt
 318     =>mj
 319     =>big5-cdp
 320     =>daikanwa
 321     =>daikanwa/ho
 322     =>cns11643-5
 323     =>cns11643-7
 324     =big5
 325     =big5-eten
 326     =>gt-k
 327     =zinbun-oracle
 328     =>zinbun-oracle
 329     =ruimoku-v6
 330     =>>ruimoku-v6
 331     ==ruimoku-v6
 332     =jef-china3
 333     =>cbeta
 334     =shinjigen
 335     =ucs-var-001
 336     =ucs-var-002
 337     =ucs-var-003
 338     =ucs-var-004
 339     =ucs-var-005
 340     =ucs-var-006
 341     =ucs-var-008
 342     =ucs-var-010
 343     =ucs-itaiji-001
 344     =ucs-itaiji-002
 345     =ucs-itaiji-003
 346     =ucs-itaiji-004
 347     =ucs-itaiji-005
 348     =ucs-itaiji-006
 349     =ucs-itaiji-007
 350     =ucs-itaiji-008
 351     =ucs-itaiji-009
 352     =ucs-itaiji-010
 353     =ucs-itaiji-011
 354     =ucs-itaiji-012
 355     =>ucs-itaiji-001
 356     =>ucs-itaiji-002
 357     =>ucs-itaiji-003
 358     =>ucs-itaiji-004
 359     =>ucs-itaiji-005
 360     =>ucs-itaiji-006
 361     =>ucs-itaiji-007
 362     =>ucs-itaiji-009
 363     =big5-cdp-var-001
 364     =big5-cdp-var-002
 365     =big5-cdp-var-003
 366     =big5-cdp-var-004
 367     =big5-cdp-var-005
 368     =big5-cdp-var-010
 369     =big5-cdp-itaiji-001
 370     =big5-cdp-itaiji-002
 371     =big5-cdp-itaiji-003
 372     =>big5-cdp-itaiji-001
 373     =>ucs@iwds-1
 374     =>ucs@cognate
 375     =>ucs@component
 376     =>ucs-itaiji-001@iwds-1
 377     =>big5-cdp@iwds-1
 378     =>big5-cdp@component
 379     =>big5-cdp@cognate
 380     ==ucs@gb
 381     =ucs@gb
 382     ==ucs-var-002
 383     =ucs@JP/hanazono))
 384
 385
 386 ;;; @ char-db formatters
 387 ;;;
 388
 389 (defun char-db-make-char-spec (char)
 390   (let (ret char-spec)
 391     (cond ((characterp char)
 392            (cond ((and (setq ret (encode-char char '=ucs 'defined-only))
 393                        (not (and (<= #xE000 ret)(<= ret #xF8FF))))
 394                   (setq char-spec (list (cons '=ucs ret)))
 395                   (cond ((setq ret (get-char-attribute char 'name))
 396                          (setq char-spec (cons (cons 'name ret) char-spec))
 397                          )
 398                         ((setq ret (get-char-attribute char 'name*))
 399                          (setq char-spec (cons (cons 'name* ret) char-spec))
 400                          ))
 401                   )
 402                  ((encode-char char '=mj 'defined-only)
 403                   (setq char-spec nil)
 404                   (dolist (ccs (charset-list))
 405                     (if (and (or (eq ccs '=mj)
 406                                  ;; (eq (charset-property ccs 'iso-ir) 177)
 407                                  (string-match "=ucs@" (symbol-name ccs))
 408                                  )
 409                              (setq ccs (charset-name ccs))
 410                              (null (assq ccs char-spec))
 411                              (setq ret (encode-char char ccs 'defined-only)))
 412                         (setq char-spec (cons (cons ccs ret) char-spec))))
 413                   )
 414                  ((encode-char char '==mj 'defined-only)
 415                   (setq char-spec nil)
 416                   (dolist (ccs (charset-list))
 417                     (if (and (or (eq ccs '==mj)
 418                                  ;; (eq (charset-property ccs 'iso-ir) 177)
 419                                  (string-match "=ucs@" (symbol-name ccs))
 420                                  )
 421                              (setq ccs (charset-name ccs))
 422                              (null (assq ccs char-spec))
 423                              (setq ret (encode-char char ccs 'defined-only)))
 424                         (setq char-spec (cons (cons ccs ret) char-spec))))
 425                   )
 426                  ((encode-char char '=adobe-japan1 'defined-only)
 427                   (setq char-spec nil)
 428                   (dolist (ccs (charset-list))
 429                     (if (and (or (memq ccs
 430                                        '(=adobe-japan1-0
 431                                          =adobe-japan1-1
 432                                          =adobe-japan1-2
 433                                          =adobe-japan1-3
 434                                          =adobe-japan1-4
 435                                          =adobe-japan1-5
 436                                          =adobe-japan1-6
 437                                          ))
 438                                  ;; (eq (charset-property ccs 'iso-ir) 177)
 439                                  (string-match "=ucs@" (symbol-name ccs))
 440                                  )
 441                              (setq ccs (charset-name ccs))
 442                              (null (assq ccs char-spec))
 443                              (setq ret (encode-char char ccs 'defined-only)))
 444                         (setq char-spec (cons (cons ccs ret) char-spec))))
 445                   )
 446                  ((encode-char char '==adobe-japan1 'defined-only)
 447                   (setq char-spec nil)
 448                   (dolist (ccs (charset-list))
 449                     (if (and (or (memq ccs
 450                                        '(==adobe-japan1-0
 451                                          ==adobe-japan1-1
 452                                          ==adobe-japan1-2
 453                                          ==adobe-japan1-3
 454                                          ==adobe-japan1-4
 455                                          ==adobe-japan1-5
 456                                          ==adobe-japan1-6
 457                                          ))
 458                                  ;; (eq (charset-property ccs 'iso-ir) 177)
 459                                  (string-match "=ucs@" (symbol-name ccs))
 460                                  )
 461                              (setq ccs (charset-name ccs))
 462                              (null (assq ccs char-spec))
 463                              (setq ret (encode-char char ccs 'defined-only)))
 464                         (setq char-spec (cons (cons ccs ret) char-spec))))
 465                   )
 466                  ((setq ret
 467                         (catch 'tag
 468                           (let ((rest char-db-coded-charset-priority-list)
 469                                 ccs)
 470                             (while rest
 471                               (setq ccs (charset-name
 472                                          (find-charset (car rest))))
 473                               (if (setq ret
 474                                         (encode-char char ccs
 475                                                      'defined-only))
 476                                   (throw 'tag (cons ccs ret)))
 477                               (setq rest (cdr rest))))))
 478                   (setq char-spec (list ret))
 479                   (dolist (ccs (delq (car ret) (charset-list)))
 480                     (if (and (or (charset-iso-final-char ccs)
 481                                  (memq ccs
 482                                        '(=daikanwa
 483                                          =daikanwa@rev2
 484                                          ;; =gt-k
 485                                          =jis-x0208@1997
 486                                          ))
 487                                  (and (eq (charset-property ccs 'iso-ir) 177)
 488                                       (not (eq (charset-name ccs) '=ucs@big5)))
 489                                  ;; (string-match "=ucs@" (symbol-name ccs))
 490                                  )
 491                              (setq ccs (charset-name ccs))
 492                              (null (assq ccs char-spec))
 493                              (setq ret (encode-char char ccs 'defined-only)))
 494                         (setq char-spec (cons (cons ccs ret) char-spec))))
 495                   (if (null char-spec)
 496                       (setq char-spec (split-char char)))
 497                   (cond ((setq ret (get-char-attribute char 'name))
 498                          (setq char-spec (cons (cons 'name ret) char-spec))
 499                          )
 500                         ((setq ret (get-char-attribute char 'name*))
 501                          (setq char-spec (cons (cons 'name* ret) char-spec))
 502                          ))
 503                   )
 504                  ((setq ret (get-char-attribute
 505                              char 'ideographic-combination))
 506                   (setq char-spec
 507                         (cons (cons 'ideographic-combination ret)
 508                               char-spec))
 509                   ))
 510            char-spec)
 511           ((consp char)
 512            char))))
 513
 514 (defun char-db-insert-char-spec (char &optional readable column
 515                                       required-features)
 516   (unless column
 517     (setq column (current-column)))
 518   (let (char-spec temp-char)
 519     (setq char-spec (char-db-make-char-spec char))
 520     (unless (or (characterp char) ; char
 521                 (condition-case nil
 522                     (setq char (find-char char-spec))
 523                   (error nil)))
 524       ;; define temporary character
 525       ;;   Current implementation is dirty.
 526       (setq temp-char (define-char (cons '(ideograph-daikanwa . 0)
 527                                          char-spec)))
 528       (remove-char-attribute temp-char 'ideograph-daikanwa)
 529       (setq char temp-char))
 530     (insert-char-attributes char
 531                             readable
 532                             (union (mapcar #'car char-spec)
 533                                    required-features)
 534                             nil 'for-sub-node)
 535     (when temp-char
 536       ;; undefine temporary character
 537       ;;   Current implementation is dirty.
 538       (setq char-spec (char-attribute-alist temp-char))
 539       (while char-spec
 540         (remove-char-attribute temp-char (car (car char-spec)))
 541         (setq char-spec (cdr char-spec))))))
 542
 543 (defun char-db-insert-alist (alist &optional readable column)
 544   (unless column
 545     (setq column (current-column)))
 546   (let ((line-breaking
 547          (concat "\n" (make-string (1+ column) ?\ )))
 548         name value
 549         ret al ; cal
 550         key
 551         lbs cell rest separator)
 552     (insert "(")
 553     (while alist
 554       (setq name (car (car alist))
 555             value (cdr (car alist)))
 556       (cond ((eq name 'char)
 557              (insert "(char . ")
 558              (if (setq ret (condition-case nil
 559                                (find-char value)
 560                              (error nil)))
 561                  (progn
 562                    (setq al nil
 563                          ;; cal nil
 564                          )
 565                    (while value
 566                      (setq key (car (car value)))
 567                      ;; (if (find-charset key)
 568                      ;;     (setq cal (cons key cal))
 569                      (setq al (cons key al))
 570                      ;; )
 571                      (setq value (cdr value)))
 572                    (insert-char-attributes ret
 573                                            readable
 574                                            (or al 'none) ; cal
 575                                            nil 'for-sub-node))
 576                (insert (prin1-to-string value)))
 577              (insert ")")
 578              (insert line-breaking))
 579             ((consp value)
 580              (insert (format "(%-18s " name))
 581              (setq lbs (concat "\n" (make-string (current-column) ?\ )))
 582              (while (consp value)
 583                (setq cell (car value))
 584                (if (and (consp cell)
 585                         (consp (car cell))
 586                         (setq ret (condition-case nil
 587                                       (find-char cell)
 588                                     (error nil)))
 589                         )
 590                    (progn
 591                      (setq rest cell
 592                            al nil
 593                            ;; cal nil
 594                            )
 595                      (while rest
 596                        (setq key (car (car rest)))
 597                        ;; (if (find-charset key)
 598                        ;;     (setq cal (cons key cal))
 599                        (setq al (cons key al))
 600                        ;; )
 601                        (setq rest (cdr rest)))
 602                      (if separator
 603                          (insert lbs))
 604                      (insert-char-attributes ret
 605                                              readable
 606                                              al ; cal
 607                                              nil 'for-sub-node)
 608                      (setq separator lbs))
 609                  (if separator
 610                      (insert separator))
 611                  (insert (prin1-to-string cell))
 612                  (setq separator " "))
 613                (setq value (cdr value)))
 614              (insert ")")
 615              (insert line-breaking))
 616             (t
 617              (insert (format "(%-18s . %S)%s"
 618                              name value
 619                              line-breaking))))
 620       (setq alist (cdr alist))))
 621   (insert ")"))
 622
 623 (defun char-db-insert-char-reference (plist &optional readable column)
 624   (unless column
 625     (setq column (current-column)))
 626   (let ((line-breaking
 627          (concat "\n" (make-string (1+ column) ?\ )))
 628         (separator "")
 629         name value)
 630     (insert "(")
 631     (while plist
 632       (setq name (pop plist))
 633       (setq value (pop plist))
 634       (cond ((eq name :char)
 635              (insert separator)
 636              (insert ":char\t")
 637              (cond ((numberp value)
 638                     (setq value (decode-char '=ucs value)))
 639                    ;; ((consp value)
 640                    ;;  (setq value (or (find-char value)
 641                    ;;                  value)))
 642                    )
 643              (char-db-insert-char-spec value readable)
 644              (insert line-breaking)
 645              (setq separator ""))
 646             ((eq name :radical)
 647              (insert (format "%s%s\t%d ; %c%s"
 648                              separator
 649                              name value
 650                              (ideographic-radical value)
 651                              line-breaking))
 652              (setq separator ""))
 653             (t
 654              (insert (format "%s%s\t%S" separator name value))
 655              (setq separator line-breaking)))
 656       ))
 657   (insert ")"))
 658
 659 (defun char-db-decode-isolated-char (ccs code-point)
 660   (let (ret)
 661     (setq ret
 662           (cond ((eq ccs 'arabic-iso8859-6)
 663                  (decode-char ccs code-point))
 664                 ;; ((eq ccs '=gt)
 665                 ;;  (decode-builtin-char '==gt code-point))
 666                 ((and (memq ccs '(=gt-pj-1
 667                                   =gt-pj-2
 668                                   =gt-pj-3
 669                                   =gt-pj-4
 670                                   =gt-pj-5
 671                                   =gt-pj-6
 672                                   =gt-pj-7
 673                                   =gt-pj-8
 674                                   =gt-pj-9
 675                                   =gt-pj-10
 676                                   =gt-pj-11))
 677                       (setq ret (decode-char ccs code-point))
 678                       (setq ret (encode-char ret '=gt 'defined-only)))
 679                  (decode-builtin-char '=gt ret))
 680                 (t
 681                  (decode-builtin-char ccs code-point))))
 682     (cond ((null ret)
 683            (or (decode-char ccs code-point)
 684                (define-char (list (cons ccs code-point)))))
 685           ((and (<= 0 (char-int ret))
 686                 (<= (char-int ret) #x1F))
 687            (decode-char '=ucs (+ #x2400 (char-int ret))))
 688           ((= (char-int ret) #x7F)
 689            ?\u2421)
 690           (t ret))))
 691
 692 (defvar char-db-convert-obsolete-format t)
 693
 694 (defun char-db-insert-ccs-feature (name value line-breaking)
 695   (cond
 696    ((integerp value)
 697     (insert
 698      (format
 699       (cond
 700        ((memq name '(=>iwds-1
 701                      ==shinjigen
 702                      =shinjigen
 703                      =shinjigen@1ed ==shinjigen@1ed
 704                      =shinjigen@rev ==shinjigen@rev
 705                      =shinjigen/+p@rev ==shinjigen/+p@rev
 706                      ===daikanwa/ho ==daikanwa/ho
 707                      =daikanwa/ho =>>daikanwa/ho =>daikanwa/ho))
 708         "(%-18s .  %04d)\t; %c")
 709        ((eq name '=shinjigen@1ed/24pr)
 710         "(%-18s . %04d)\t; %c")
 711        ((or
 712          (memq name
 713                '(===daikanwa
 714                  ==daikanwa =daikanwa =>>daikanwa =>daikanwa
 715                  =daikanwa@rev1 =daikanwa@rev2
 716                  =daikanwa/+p ==daikanwa/+p ===daikanwa/+p
 717                  =>>daikanwa/+p
 718                  =daikanwa/+2p =>>daikanwa/+2p
 719                  =gt ==gt ===gt
 720                  =>>gt =+>gt =>gt
 721                  =gt-k ==gt-k ===gt-k
 722                  =>>gt-k =>gt-k
 723                  =adobe-japan1-0 ==adobe-japan1-0 ===adobe-japan1-0
 724                  =adobe-japan1-1 ==adobe-japan1-1 ===adobe-japan1-1
 725                  =adobe-japan1-2 ==adobe-japan1-2 ===adobe-japan1-2
 726                  =adobe-japan1-3 ==adobe-japan1-3 ===adobe-japan1-3
 727                  =adobe-japan1-4 ==adobe-japan1-4 ===adobe-japan1-4
 728                  =adobe-japan1-5 ==adobe-japan1-5 ===adobe-japan1-5
 729                  =adobe-japan1-6 ==adobe-japan1-6 ===adobe-japan1-6
 730                  =>>adobe-japan1-0 =+>adobe-japan1-0
 731                  =>>adobe-japan1-1 =+>adobe-japan1-1
 732                  =>>adobe-japan1-2 =+>adobe-japan1-2
 733                  =>>adobe-japan1-3 =+>adobe-japan1-3
 734                  =>>adobe-japan1-4 =+>adobe-japan1-4
 735                  =>>adobe-japan1-5 =+>adobe-japan1-5
 736                  =>>adobe-japan1-6 =+>adobe-japan1-6
 737                  =>cbeta =cbeta =>>cbeta ==cbeta ===cbeta
 738                  =zinbun-oracle =>zinbun-oracle
 739                  =daijiten ==daijiten ===daijiten ==>daijiten
 740                  ===hng-jou ===hng-keg ===hng-dng ===hng-mam
 741                  ===hng-drt ===hng-kgk ===hng-myz ===hng-kda
 742                  ===hng-khi ===hng-khm ===hng-fhs ===hng-hok
 743                  ===hng-kyd ===hng-sok
 744                  ===hng-yhk ===hng-kak ===hng-kar ===hng-kae
 745                  ===hng-sys ===hng-tsu ===hng-tzj
 746                  ===hng-hos ===hng-nak ===hng-jhk
 747                  ===hng-hod ===hng-gok ===hng-ink ===hng-nto
 748                  ===hng-nkm ===hng-k24 ===hng-nkk
 749                  ===hng-kcc ===hng-kcj ===hng-kbk ===hng-sik
 750                  ===hng-skk ===hng-kyu ===hng-ksk ===hng-wan
 751                  ===hng-okd ===hng-wad ===hng-kmi
 752                  ===hng-zkd ===hng-doh ===hng-jyu
 753                  ===hng-tzs ===hng-kss ===hng-kyo
 754                  ===hng-smk
 755                  =shuowen-jiguge ===shuowen-jiguge4 ===shuowen-jiguge5))
 756          ;; (string-match "^=adobe-" (symbol-name name))
 757          )
 758         "(%-18s . %05d)\t; %c")
 759        ((memq name '(=hanyo-denshi/ks
 760                      ==hanyo-denshi/ks ===hanyo-denshi/ks
 761                      =>>hanyo-denshi/ks
 762                      =koseki ==koseki
 763                      =mj ==mj ===mj =>>mj =>mj
 764                      =zihai mojikyo))
 765         "(%-18s . %06d)\t; %c")
 766        ((memq name '(=hanyo-denshi/tk ==hanyo-denshi/tk))
 767         "(%-18s . %08d)\t; %c")
 768        ((>= (charset-dimension name) 2)
 769         "(%-18s . #x%04X)\t; %c")
 770        (t
 771         "(%-18s . #x%02X)\t; %c"))
 772       name
 773       (if (= (charset-iso-graphic-plane name) 1)
 774           (logior value
 775                   (cond ((= (charset-dimension name) 1)
 776                          #x80)
 777                         ((= (charset-dimension name) 2)
 778                          #x8080)
 779                         ((= (charset-dimension name) 3)
 780                          #x808080)
 781                         (t 0)))
 782         value)
 783       (char-db-decode-isolated-char name value)))
 784     (if (and (= (charset-chars name) 94)
 785              (= (charset-dimension name) 2))
 786         (insert (format " [%02d-%02d]"
 787                         (- (lsh value -8) 32)
 788                         (- (logand value 255) 32))))
 789     )
 790    (t
 791     (insert (format "(%-18s . %s)" name value))
 792     ))
 793   (insert line-breaking))
 794
 795 (defun char-db-insert-relation-feature (char name value line-breaking
 796                                              ccss readable)
 797   (insert (format "(%-18s%s " name line-breaking))
 798   (let ((lbs (concat "\n" (make-string (current-column) ?\ )))
 799         separator cell sources required-features
 800         ret)
 801     (while (consp value)
 802       (setq cell (car value))
 803       (if (integerp cell)
 804           (setq cell (decode-char '=ucs cell)))
 805       (cond
 806        ((eq name '->subsumptive)
 807         (when (or (not (some (lambda (atr)
 808                                (get-char-attribute cell atr))
 809                              char-db-ignored-attributes))
 810                   (some (lambda (ccs)
 811                           (encode-char cell ccs 'defined-only))
 812                         ccss))
 813           (if separator
 814               (insert lbs))
 815           (let ((char-db-ignored-attributes
 816                  (cons '<-subsumptive
 817                        char-db-ignored-attributes)))
 818             (insert-char-attributes cell readable nil nil 'for-sub-node))
 819           (setq separator lbs))
 820         )
 821        ((characterp cell)
 822         (setq sources
 823               (get-char-attribute
 824                char (intern (format "%s*sources" name))))
 825         (setq required-features nil)
 826         (dolist (source sources)
 827           (cond
 828            ((memq source '(JP
 829                            JP/Jouyou
 830                            shinjigen shinjigen@1ed shinjigen@rev))
 831             (setq required-features
 832                   (union required-features
 833                          '(=jis-x0208
 834                            =jis-x0208@1990
 835                            =jis-x0213-1@2000
 836                            =jis-x0213-1@2004
 837                            =jis-x0213-2
 838                            =jis-x0212
 839                            =jis-x0208@1983
 840                            =jis-x0208@1978
 841                            =shinjigen
 842                            =shinjigen@1ed
 843                            =shinjigen@rev
 844                            =shinjigen/+p@rev))))
 845            ((eq source 'CN)
 846             (setq required-features
 847                   (union required-features
 848                          '(=gb2312
 849                            =gb12345
 850                            =iso-ir165)))))
 851           (cond
 852            ((find-charset (setq ret (intern (format "=%s" source))))
 853             (setq required-features
 854                   (cons ret required-features)))
 855            (t (setq required-features
 856                     (cons source required-features)))))
 857         (cond ((string-match "@JP" (symbol-name name))
 858                (setq required-features
 859                      (union required-features
 860                             '(=jis-x0208
 861                               =jis-x0208@1990
 862                               =jis-x0213-1-2000
 863                               =jis-x0213-2-2000
 864                               =jis-x0212
 865                               =jis-x0208@1983
 866                               =jis-x0208@1978))))
 867               ((string-match "@CN" (symbol-name name))
 868                (setq required-features
 869                      (union required-features
 870                             '(=gb2312
 871                               =gb12345
 872                               =iso-ir165)))))
 873         (if separator
 874             (insert lbs))
 875         (if readable
 876             (insert (format "%S" cell))
 877           (char-db-insert-char-spec cell readable
 878                                     nil
 879                                     required-features))
 880         (setq separator lbs))
 881        ((consp cell)
 882         (if separator
 883             (insert lbs))
 884         (if (consp (car cell))
 885             (char-db-insert-char-spec cell readable)
 886           (char-db-insert-char-reference cell readable))
 887         (setq separator lbs))
 888        (t
 889         (if separator
 890             (insert separator))
 891         (insert (prin1-to-string cell))
 892         (setq separator " ")))
 893       (setq value (cdr value)))
 894     (insert ")")
 895     (insert line-breaking)))
 896
 897 (defun insert-char-attributes (char &optional readable attributes column
 898                                     for-sub-node)
 899   (unless column
 900     (setq column (current-column)))
 901   (let (name value ; has-long-ccs-name
 902         rest
 903         radical strokes
 904         (line-breaking
 905          (concat "\n" (make-string (1+ column) ?\ )))
 906         lbs cell separator ret
 907         key al cal
 908         dest-ccss ; sources required-features
 909         ccss)
 910     (let (atr-d)
 911       (setq attributes
 912             (sort (if attributes
 913                       (if (consp attributes)
 914                           (progn
 915                             (dolist (name attributes)
 916                               (unless (memq name char-db-ignored-attributes)
 917                                 (if (find-charset name)
 918                                     (push name ccss))
 919                                 (push name atr-d)))
 920                             atr-d))
 921                     (dolist (name (char-attribute-list))
 922                       (unless (memq name char-db-ignored-attributes)
 923                         (if (find-charset name)
 924                             (push name ccss))
 925                         (push name atr-d)))
 926                     atr-d)
 927                   #'char-attribute-name<)))
 928     (insert "(")
 929     (when (memq '<-subsumptive attributes)
 930       (when (or readable (not for-sub-node))
 931         (when (setq value (get-char-attribute char '<-subsumptive))
 932           (char-db-insert-relation-feature char '<-subsumptive value
 933                                            line-breaking
 934                                            ccss readable)))
 935       (setq attributes (delq '<-subsumptive attributes)))
 936     (when (and (memq '<-denotational attributes)
 937                (setq value (get-char-attribute char '<-denotational)))
 938       (char-db-insert-relation-feature char '<-denotational value
 939                                        line-breaking
 940                                        ccss readable)
 941       (setq attributes (delq '<-denotational attributes)))
 942     (when (and (memq '<-denotational@component attributes)
 943                (setq value (get-char-attribute char '<-denotational@component)))
 944       (char-db-insert-relation-feature char '<-denotational@component value
 945                                        line-breaking
 946                                        ccss readable)
 947       (setq attributes (delq '<-denotational@component attributes)))
 948     (when (and (memq '<-denotational@usage attributes)
 949                (setq value (get-char-attribute char '<-denotational@usage)))
 950       (char-db-insert-relation-feature char '<-denotational@usage value
 951                                        line-breaking
 952                                        ccss readable)
 953       (setq attributes (delq '<-denotational@usage attributes)))
 954     (when (and (memq 'name attributes)
 955                (setq value (get-char-attribute char 'name)))
 956       (insert (format
 957                (if (> (+ (current-column) (length value)) 48)
 958                    "(name . %S)%s"
 959                  "(name               . %S)%s")
 960                value line-breaking))
 961       (setq attributes (delq 'name attributes))
 962       )
 963     (when (and (memq 'name* attributes)
 964                (setq value (get-char-attribute char 'name*)))
 965       (insert (format
 966                (if (> (+ (current-column) (length value)) 48)
 967                    "(name* . %S)%s"
 968                  "(name*              . %S)%s")
 969                value line-breaking))
 970       (setq attributes (delq 'name* attributes))
 971       )
 972     (when (and (memq 'script attributes)
 973                (setq value (get-char-attribute char 'script)))
 974       (insert (format "(script\t\t%s)%s"
 975                       (mapconcat (function prin1-to-string)
 976                                  value " ")
 977                       line-breaking))
 978       (setq attributes (delq 'script attributes))
 979       )
 980     (dolist (name '(=>ucs =>ucs*))
 981       (when (and (memq name attributes)
 982                  (setq value (get-char-attribute char name)))
 983         (insert (format "(%-18s . #x%04X)\t; %c%s"
 984                         name value (decode-char '=ucs value)
 985                         line-breaking))
 986         (setq attributes (delq name attributes))))
 987     (dolist (name '(=>ucs@gb =>ucs@big5))
 988       (when (and (memq name attributes)
 989                  (setq value (get-char-attribute char name)))
 990         (insert (format "(%-18s . #x%04X)\t; %c%s"
 991                         name value
 992                         (decode-char (intern
 993                                       (concat "="
 994                                               (substring
 995                                                (symbol-name name) 2)))
 996                                      value)
 997                         line-breaking))
 998         (setq attributes (delq name attributes))
 999         ))
1000     ;; (dolist (name '(=>daikanwa))
1001     ;;   (when (and (memq name attributes)
1002     ;;              (setq value (get-char-attribute char name)))
1003     ;;     (insert
1004     ;;      (if (integerp value)
1005     ;;          (format "(%-18s . %05d)\t; %c%s"
1006     ;;                  name value (decode-char '=daikanwa value)
1007     ;;                  line-breaking)
1008     ;;        (format "(%-18s %s)\t; %c%s"
1009     ;;                name
1010     ;;                (mapconcat (function prin1-to-string)
1011     ;;                           value " ")
1012     ;;                (char-representative-of-daikanwa char)
1013     ;;                line-breaking)))
1014     ;;     (setq attributes (delq name attributes))))
1015     (when (and (memq 'general-category attributes)
1016                (setq value (get-char-attribute char 'general-category)))
1017       (insert (format
1018                "(general-category\t%s) ; %s%s"
1019                (mapconcat (lambda (cell)
1020                             (format "%S" cell))
1021                           value " ")
1022                (cond ((rassoc value unidata-normative-category-alist)
1023                       "Normative Category")
1024                      ((rassoc value unidata-informative-category-alist)
1025                       "Informative Category")
1026                      (t
1027                       "Unknown Category"))
1028                line-breaking))
1029       (setq attributes (delq 'general-category attributes))
1030       )
1031     (when (and (memq 'bidi-category attributes)
1032                (setq value (get-char-attribute char 'bidi-category)))
1033       (insert (format "(bidi-category\t. %S)%s"
1034                       value
1035                       line-breaking))
1036       (setq attributes (delq 'bidi-category attributes))
1037       )
1038     (unless (or (not (memq 'mirrored attributes))
1039                 (eq (setq value (get-char-attribute char 'mirrored 'empty))
1040                     'empty))
1041       (insert (format "(mirrored\t\t. %S)%s"
1042                       value
1043                       line-breaking))
1044       (setq attributes (delq 'mirrored attributes))
1045       )
1046     (cond
1047      ((and (memq 'decimal-digit-value attributes)
1048            (setq value (get-char-attribute char 'decimal-digit-value)))
1049       (insert (format "(decimal-digit-value . %S)%s"
1050                       value
1051                       line-breaking))
1052       (setq attributes (delq 'decimal-digit-value attributes))
1053       (when (and (memq 'digit-value attributes)
1054                  (setq value (get-char-attribute char 'digit-value)))
1055         (insert (format "(digit-value\t . %S)%s"
1056                         value
1057                         line-breaking))
1058         (setq attributes (delq 'digit-value attributes))
1059         )
1060       (when (and (memq 'numeric-value attributes)
1061                  (setq value (get-char-attribute char 'numeric-value)))
1062         (insert (format "(numeric-value\t . %S)%s"
1063                         value
1064                         line-breaking))
1065         (setq attributes (delq 'numeric-value attributes))
1066         )
1067       )
1068      (t
1069       (when (and (memq 'digit-value attributes)
1070                  (setq value (get-char-attribute char 'digit-value)))
1071         (insert (format "(digit-value\t. %S)%s"
1072                         value
1073                         line-breaking))
1074         (setq attributes (delq 'digit-value attributes))
1075         )
1076       (when (and (memq 'numeric-value attributes)
1077                  (setq value (get-char-attribute char 'numeric-value)))
1078         (insert (format "(numeric-value\t. %S)%s"
1079                         value
1080                         line-breaking))
1081         (setq attributes (delq 'numeric-value attributes))
1082         )))
1083     (when (and (memq 'iso-10646-comment attributes)
1084                (setq value (get-char-attribute char 'iso-10646-comment)))
1085       (insert (format "(iso-10646-comment\t. %S)%s"
1086                       value
1087                       line-breaking))
1088       (setq attributes (delq 'iso-10646-comment attributes))
1089       )
1090     (when (and (memq 'morohashi-daikanwa attributes)
1091                (setq value (get-char-attribute char 'morohashi-daikanwa)))
1092       (insert (format "(morohashi-daikanwa\t%s)%s"
1093                       (mapconcat (function prin1-to-string) value " ")
1094                       line-breaking))
1095       (setq attributes (delq 'morohashi-daikanwa attributes))
1096       )
1097     (setq radical nil
1098           strokes nil)
1099     (when (and (memq 'ideographic-radical attributes)
1100                (setq value (get-char-attribute char 'ideographic-radical)))
1101       (setq radical value)
1102       (insert (format "(ideographic-radical . %S)\t; %c%s"
1103                       radical
1104                       (ideographic-radical radical)
1105                       line-breaking))
1106       (setq attributes (delq 'ideographic-radical attributes))
1107       )
1108     (when (and (memq 'shuowen-radical attributes)
1109                (setq value (get-char-attribute char 'shuowen-radical)))
1110       (insert (format "(shuowen-radical\t. %S)\t; %c%s"
1111                       value
1112                       (shuowen-radical value)
1113                       line-breaking))
1114       (setq attributes (delq 'shuowen-radical attributes))
1115       )
1116     (let (key)
1117       (dolist (domain
1118                (append
1119                 char-db-feature-domains
1120                 (let (dest domain)
1121                   (dolist (feature (char-attribute-list))
1122                     (setq feature (symbol-name feature))
1123                     (when (string-match
1124                            "\\(radical\\|strokes\\)@\\([^@*]+\\)\\(\\*\\|$\\)"
1125                            feature)
1126                       (setq domain (intern (match-string 2 feature)))
1127                      (unless (memq domain dest)
1128                        (setq dest (cons domain dest)))))
1129                   (sort dest #'string<))))
1130         (setq key (intern (format "%s@%s" 'ideographic-radical domain)))
1131         (when (and (memq key attributes)
1132                    (setq value (get-char-attribute char key)))
1133           (setq radical value)
1134           (insert (format "(%s . %S)\t; %c%s"
1135                           key
1136                           radical
1137                           (ideographic-radical radical)
1138                           line-breaking))
1139           (setq attributes (delq key attributes))
1140           )
1141         (setq key (intern (format "%s@%s" 'ideographic-strokes domain)))
1142         (when (and (memq key attributes)
1143                    (setq value (get-char-attribute char key)))
1144           (setq strokes value)
1145           (insert (format "(%s . %S)%s"
1146                           key
1147                           strokes
1148                           line-breaking))
1149           (setq attributes (delq key attributes))
1150           )
1151         (setq key (intern (format "%s@%s" 'total-strokes domain)))
1152         (when (and (memq key attributes)
1153                    (setq value (get-char-attribute char key)))
1154           (insert (format "(%s       . %S)%s"
1155                           key
1156                           value
1157                           line-breaking))
1158           (setq attributes (delq key attributes))
1159           )
1160         (dolist (feature '(ideographic-radical
1161                            ideographic-strokes
1162                            total-strokes))
1163           (setq key (intern (format "%s@%s*sources" feature domain)))
1164           (when (and (memq key attributes)
1165                      (setq value (get-char-attribute char key)))
1166             (insert (format "(%s%s" key line-breaking))
1167             (dolist (cell value)
1168               (insert (format " %s" cell)))
1169             (insert ")")
1170             (insert line-breaking)
1171             (setq attributes (delq key attributes))
1172             ))
1173         ))
1174     (when (and (memq 'ideographic-strokes attributes)
1175                (setq value (get-char-attribute char 'ideographic-strokes)))
1176       (setq strokes value)
1177       (insert (format "(ideographic-strokes . %S)%s"
1178                       strokes
1179                       line-breaking))
1180       (setq attributes (delq 'ideographic-strokes attributes))
1181       )
1182     (when (and (memq 'kangxi-radical attributes)
1183                (setq value (get-char-attribute char 'kangxi-radical)))
1184       (unless (eq value radical)
1185         (insert (format "(kangxi-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 'kangxi-radical attributes))
1192       )
1193     (when (and (memq 'kangxi-strokes attributes)
1194                (setq value (get-char-attribute char 'kangxi-strokes)))
1195       (unless (eq value strokes)
1196         (insert (format "(kangxi-strokes\t . %S)%s"
1197                         value
1198                         line-breaking))
1199         (or strokes
1200             (setq strokes value)))
1201       (setq attributes (delq 'kangxi-strokes attributes))
1202       )
1203     ;; (when (and (memq 'japanese-radical attributes)
1204     ;;            (setq value (get-char-attribute char 'japanese-radical)))
1205     ;;   (unless (eq value radical)
1206     ;;     (insert (format "(japanese-radical\t . %S)\t; %c%s"
1207     ;;                     value
1208     ;;                     (ideographic-radical value)
1209     ;;                     line-breaking))
1210     ;;     (or radical
1211     ;;         (setq radical value)))
1212     ;;   (setq attributes (delq 'japanese-radical attributes))
1213     ;;   )
1214     (when (and (memq 'japanese-strokes attributes)
1215                (setq value (get-char-attribute char 'japanese-strokes)))
1216       (unless (eq value strokes)
1217         (insert (format "(japanese-strokes\t . %S)%s"
1218                         value
1219                         line-breaking))
1220         (or strokes
1221             (setq strokes value)))
1222       (setq attributes (delq 'japanese-strokes attributes))
1223       )
1224     (when (and (memq 'cns-radical attributes)
1225                (setq value (get-char-attribute char 'cns-radical)))
1226       (insert (format "(cns-radical\t . %S)\t; %c%s"
1227                       value
1228                       (ideographic-radical value)
1229                       line-breaking))
1230       (setq attributes (delq 'cns-radical attributes))
1231       )
1232     (when (and (memq 'cns-strokes attributes)
1233                (setq value (get-char-attribute char 'cns-strokes)))
1234       (unless (eq value strokes)
1235         (insert (format "(cns-strokes\t . %S)%s"
1236                         value
1237                         line-breaking))
1238         (or strokes
1239             (setq strokes value)))
1240       (setq attributes (delq 'cns-strokes attributes))
1241       )
1242     ;; (when (and (memq 'ideographic- attributes)
1243     ;;            (setq value (get-char-attribute char 'ideographic-)))
1244     ;;   (insert "(ideographic-       ")
1245     ;;   (setq lbs (concat "\n" (make-string (current-column) ?\ ))
1246     ;;         separator nil)
1247     ;;   (while (consp value)
1248     ;;     (setq cell (car value))
1249     ;;     (if (integerp cell)
1250     ;;         (setq cell (decode-char '=ucs cell)))
1251     ;;     (cond ((characterp cell)
1252     ;;            (if separator
1253     ;;                (insert lbs))
1254     ;;            (if readable
1255     ;;                (insert (format "%S" cell))
1256     ;;              (char-db-insert-char-spec cell readable))
1257     ;;            (setq separator lbs))
1258     ;;           ((consp cell)
1259     ;;            (if separator
1260     ;;                (insert lbs))
1261     ;;            (if (consp (car cell))
1262     ;;                (char-db-insert-char-spec cell readable)
1263     ;;              (char-db-insert-char-reference cell readable))
1264     ;;            (setq separator lbs))
1265     ;;           (t
1266     ;;            (if separator
1267     ;;                (insert separator))
1268     ;;            (insert (prin1-to-string cell))
1269     ;;            (setq separator " ")))
1270     ;;     (setq value (cdr value)))
1271     ;;   (insert ")")
1272     ;;   (insert line-breaking)
1273     ;;   (setq attributes (delq 'ideographic- attributes)))
1274     (when (and (memq 'total-strokes attributes)
1275                (setq value (get-char-attribute char 'total-strokes)))
1276       (insert (format "(total-strokes       . %S)%s"
1277                       value
1278                       line-breaking))
1279       (setq attributes (delq 'total-strokes attributes))
1280       )
1281     ;; (when (and (memq '->ideograph attributes)
1282     ;;            (setq value (get-char-attribute char '->ideograph)))
1283     ;;   (insert (format "(->ideograph\t%s)%s"
1284     ;;                   (mapconcat (lambda (code)
1285     ;;                                (cond ((symbolp code)
1286     ;;                                       (symbol-name code))
1287     ;;                                      ((integerp code)
1288     ;;                                       (format "#x%04X" code))
1289     ;;                                      (t
1290     ;;                                       (format "%s %S"
1291     ;;                                               line-breaking code))))
1292     ;;                              value " ")
1293     ;;                   line-breaking))
1294     ;;   (setq attributes (delq '->ideograph attributes))
1295     ;;   )
1296     ;; (when (and (memq '->decomposition attributes)
1297     ;;            (setq value (get-char-attribute char '->decomposition)))
1298     ;;   (insert (format "(->decomposition\t%s)%s"
1299     ;;                   (mapconcat (lambda (code)
1300     ;;                                (cond ((symbolp code)
1301     ;;                                       (symbol-name code))
1302     ;;                                      ((characterp code)
1303     ;;                                       (if readable
1304     ;;                                           (format "%S" code)
1305     ;;                                         (format "#x%04X"
1306     ;;                                                 (char-int code))
1307     ;;                                         ))
1308     ;;                                      ((integerp code)
1309     ;;                                       (format "#x%04X" code))
1310     ;;                                      (t
1311     ;;                                       (format "%s%S" line-breaking code))))
1312     ;;                              value " ")
1313     ;;                   line-breaking))
1314     ;;   (setq attributes (delq '->decomposition attributes))
1315     ;;   )
1316     (if (equal (get-char-attribute char '->titlecase)
1317                (get-char-attribute char '->uppercase))
1318         (setq attributes (delq '->titlecase attributes)))
1319     (when (and (memq '->mojikyo attributes)
1320                (setq value (get-char-attribute char '->mojikyo)))
1321       (insert (format "(->mojikyo\t\t. %06d)\t; %c%s"
1322                       value (decode-char 'mojikyo value)
1323                       line-breaking))
1324       (setq attributes (delq '->mojikyo attributes))
1325       )
1326     (unless readable
1327       (dolist (ignored '(composition
1328                          ->denotational <-subsumptive ->ucs-unified
1329                          ;; ->ideographic-component-forms
1330                          ))
1331         (setq attributes (delq ignored attributes))))
1332     (while attributes
1333       (setq name (car attributes))
1334       (unless (eq (setq value (get-char-attribute char name 'value-is-empty))
1335                   'value-is-empty)
1336         (cond ((setq ret (find-charset name))
1337                (setq name (charset-name ret))
1338                (when (not (memq name dest-ccss))
1339                  (setq dest-ccss (cons name dest-ccss))
1340                  (char-db-insert-ccs-feature name value line-breaking))
1341                )
1342               ((string-match "^=>ucs@" (symbol-name name))
1343                (insert (format "(%-18s . #x%04X)\t; %c%s"
1344                                name value (decode-char '=ucs value)
1345                                line-breaking))
1346                )
1347               ((eq name 'jisx0208-1978/4X)
1348                (insert (format "(%-18s . #x%04X)%s"
1349                                name value
1350                                line-breaking))
1351                )
1352               ((and
1353                 (not readable)
1354                 (not (eq name '->subsumptive))
1355                 (not (eq name '->uppercase))
1356                 (not (eq name '->lowercase))
1357                 (not (eq name '->titlecase))
1358                 (not (eq name '->canonical))
1359                 (not (eq name '->Bopomofo))
1360                 (not (eq name '->mistakable))
1361                 (not (eq name '->ideographic-variants))
1362                 (null (get-char-attribute
1363                        char (intern (format "%s*sources" name))))
1364                 (not (string-match "\\*sources$" (symbol-name name)))
1365                 (null (get-char-attribute
1366                        char (intern (format "%s*note" name))))
1367                 (not (string-match "\\*note$" (symbol-name name)))
1368                 (or (eq name '<-identical)
1369                     (eq name '<-uppercase)
1370                     (eq name '<-lowercase)
1371                     (eq name '<-titlecase)
1372                     (eq name '<-canonical)
1373                     (eq name '<-ideographic-variants)
1374                     ;; (eq name '<-synonyms)
1375                     (string-match "^<-synonyms" (symbol-name name))
1376                     (eq name '<-mistakable)
1377                     (when (string-match "^->" (symbol-name name))
1378                       (cond
1379                        ((string-match "^->fullwidth" (symbol-name name))
1380                         (not (and (consp value)
1381                                   (characterp (car value))
1382                                   (encode-char
1383                                    (car value) '=ucs 'defined-only)))
1384                         )
1385                        (t)))
1386                     ))
1387                )
1388               ((or (eq name 'ideographic-structure)
1389                    (eq name 'ideographic-combination)
1390                    ;; (eq name 'ideographic-)
1391                    (eq name '=decomposition)
1392                    (char-feature-base-name= '=decomposition name)
1393                    (char-feature-base-name= '=>decomposition name)
1394                    ;; (string-match "^=>*decomposition\\(@[^*]+\\)?$"
1395                    ;;               (symbol-name name))
1396                    (string-match "^\\(->\\|<-\\)[^*]*$" (symbol-name name))
1397                    (string-match "^\\(->\\|<-\\)[^*]*\\*sources$"
1398                                  (symbol-name name))
1399                    )
1400                (char-db-insert-relation-feature char name value
1401                                                 line-breaking
1402                                                 ccss readable))
1403               ;; ((memq name '(ideograph=
1404               ;;               original-ideograph-of
1405               ;;               ancient-ideograph-of
1406               ;;               vulgar-ideograph-of
1407               ;;               wrong-ideograph-of
1408               ;;               ;; simplified-ideograph-of
1409               ;;               ideographic-variants
1410               ;;               ;; ideographic-different-form-of
1411               ;;               ))
1412               ;;  (insert (format "(%-18s%s " name line-breaking))
1413               ;;  (setq lbs (concat "\n" (make-string (current-column) ?\ ))
1414               ;;        separator nil)
1415               ;;  (while (consp value)
1416               ;;    (setq cell (car value))
1417               ;;    (if (and (consp cell)
1418               ;;             (consp (car cell)))
1419               ;;        (progn
1420               ;;          (if separator
1421               ;;              (insert lbs))
1422               ;;          (char-db-insert-alist cell readable)
1423               ;;          (setq separator lbs))
1424               ;;      (if separator
1425               ;;          (insert separator))
1426               ;;      (insert (prin1-to-string cell))
1427               ;;      (setq separator " "))
1428               ;;    (setq value (cdr value)))
1429               ;;  (insert ")")
1430               ;;  (insert line-breaking))
1431               ((consp value)
1432                (insert (format "(%-18s " name))
1433                (setq lbs (concat "\n" (make-string (current-column) ?\ ))
1434                      separator nil)
1435                (while (consp value)
1436                  (setq cell (car value))
1437                  (if (and (consp cell)
1438                           (consp (car cell))
1439                           (setq ret (condition-case nil
1440                                         (find-char cell)
1441                                       (error nil))))
1442                      (progn
1443                        (setq rest cell
1444                              al nil
1445                              cal nil)
1446                        (while rest
1447                          (setq key (car (car rest)))
1448                          (if (find-charset key)
1449                              (setq cal (cons key cal))
1450                            (setq al (cons key al)))
1451                          (setq rest (cdr rest)))
1452                        (if separator
1453                            (insert lbs))
1454                        (insert-char-attributes ret
1455                                                readable
1456                                                al ; cal
1457                                                nil 'for-sub-node)
1458                        (setq separator lbs))
1459                    (setq ret (prin1-to-string cell))
1460                    (if separator
1461                        (if (< (+ (current-column)
1462                                  (length ret)
1463                                  (length separator))
1464                               76)
1465                            (insert separator)
1466                          (insert lbs)))
1467                    (insert ret)
1468                    (setq separator " "))
1469                  (setq value (cdr value)))
1470                (insert ")")
1471                (insert line-breaking))
1472               (t
1473                (insert (format "(%-18s" name))
1474                (setq ret (prin1-to-string value))
1475                (unless (< (+ (current-column)
1476                              (length ret)
1477                              3)
1478                           76)
1479                  (insert line-breaking))
1480                (insert " . " ret ")" line-breaking)
1481                ;; (insert (format "(%-18s . %S)%s"
1482                ;;                 name value
1483                ;;                 line-breaking))
1484                )
1485               ))
1486       (setq attributes (cdr attributes)))
1487     (insert ")")))
1488
1489 (defun insert-char-data (char &optional readable
1490                               attributes)
1491   (save-restriction
1492     (narrow-to-region (point)(point))
1493     (insert "(define-char
1494   '")
1495     (insert-char-attributes char readable attributes)
1496     (insert ")\n")
1497     (goto-char (point-min))
1498     (while (re-search-forward "[ \t]+$" nil t)
1499       (replace-match ""))
1500     ;; from tabify.
1501     (goto-char (point-min))
1502     (while (re-search-forward "[ \t][ \t][ \t]*" nil t)
1503       (let ((column (current-column))
1504             (indent-tabs-mode t))
1505         (delete-region (match-beginning 0) (point))
1506         (indent-to column)))
1507     (goto-char (point-max))
1508     ;; (tabify (point-min)(point-max))
1509     ))
1510
1511 (defun insert-char-data-with-variant (char &optional printable
1512                                            no-ucs-unified
1513                                            script excluded-script)
1514   (insert-char-data char printable)
1515   (let ((variants (char-variants char))
1516         rest
1517         variant vs ret)
1518     (setq variants (sort variants #'<))
1519     (setq rest variants)
1520     (setq variants (cons char variants))
1521     (while rest
1522       (setq variant (car rest))
1523       (unless (get-char-attribute variant '<-subsumptive)
1524         (if (and (or (null script)
1525                      (null (setq vs (get-char-attribute variant 'script)))
1526                      (memq script vs))
1527                  (or (null excluded-script)
1528                      (null (setq vs (get-char-attribute variant 'script)))
1529                      (not (memq excluded-script vs))))
1530             (unless (and no-ucs-unified (get-char-attribute variant '=ucs))
1531               (insert-char-data variant printable)
1532               (if (setq ret (char-variants variant))
1533                   (while ret
1534                     (or (memq (car ret) variants)
1535                         ;; (get-char-attribute (car ret) '<-subsumptive)
1536                         (setq rest (nconc rest (list (car ret)))))
1537                     (setq ret (cdr ret)))))))
1538       (setq rest (cdr rest)))))
1539
1540 (defun insert-char-range-data (min max &optional script excluded-script)
1541   (let ((code min)
1542         char)
1543     (while (<= code max)
1544       (setq char (decode-char '=ucs code))
1545       (if (encode-char char '=ucs 'defined-only)
1546           (insert-char-data-with-variant char nil 'no-ucs-unified
1547                                          script excluded-script))
1548       (setq code (1+ code)))))
1549
1550 (defun write-char-range-data-to-file (min max file
1551                                           &optional script excluded-script)
1552   (let ((coding-system-for-write char-db-file-coding-system))
1553     (with-temp-buffer
1554       (insert (format ";; -*- coding: %s -*-\n"
1555                       char-db-file-coding-system))
1556       (insert-char-range-data min max script excluded-script)
1557       (write-region (point-min)(point-max) file))))
1558
1559 (defvar what-character-original-window-configuration)
1560
1561 ;;;###autoload
1562 (defun what-char-definition (char)
1563   (interactive (list (char-after)))
1564   (let ((buf (get-buffer-create "*Character Description*"))
1565         (the-buf (current-buffer))
1566         (win-conf (current-window-configuration)))
1567     (pop-to-buffer buf)
1568     (make-local-variable 'what-character-original-window-configuration)
1569     (setq what-character-original-window-configuration win-conf)
1570     (setq buffer-read-only nil)
1571     (erase-buffer)
1572     (condition-case err
1573         (progn
1574           (insert-char-data-with-variant char 'printable)
1575           (unless (char-attribute-alist char)
1576             (insert (format ";; = %c\n"
1577                             (let* ((rest (split-char char))
1578                                    (ccs (pop rest))
1579                                    (code (pop rest)))
1580                               (while rest
1581                                 (setq code (logior (lsh code 8)
1582                                                    (pop rest))))
1583                               (decode-char ccs code)))))
1584           ;; (char-db-update-comment)
1585           (set-buffer-modified-p nil)
1586           (view-mode the-buf (lambda (buf)
1587                                (set-window-configuration
1588                                 what-character-original-window-configuration)
1589                                ))
1590           (goto-char (point-min)))
1591       (error (progn
1592                (set-window-configuration
1593                 what-character-original-window-configuration)
1594                (signal (car err) (cdr err)))))))
1595
1596
1597 ;;; @ end
1598 ;;;
1599
1600 (provide 'char-db-util)
1601
1602 ;;; char-db-util.el ends here