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,
   5 ;;   2020 MORIOKA Tomohiko.
   6
   7 ;; Author: MORIOKA Tomohiko <tomo@kanji.zinbun.kyoto-u.ac.jp>
   8 ;; Keywords: CHISE, Character Database, ISO/IEC 10646, UCS, Unicode, MULE.
   9
  10 ;; This file is part of XEmacs CHISE.
  11
  12 ;; XEmacs CHISE is free software; you can redistribute it and/or
  13 ;; modify it under the terms of the GNU General Public License as
  14 ;; published by the Free Software Foundation; either version 2, or (at
  15 ;; your option) any later version.
  16
  17 ;; XEmacs CHISE is distributed in the hope that it will be useful,
  18 ;; but WITHOUT ANY WARRANTY; without even the implied warranty of
  19 ;; MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the GNU
  20 ;; General Public License for more details.
  21
  22 ;; You should have received a copy of the GNU General Public License
  23 ;; along with XEmacs CHISE; see the file COPYING.  If not, write to
  24 ;; the Free Software Foundation, Inc., 59 Temple Place - Suite 330,
  25 ;; Boston, MA 02111-1307, USA.
  26
  27 ;;; Code:
  28
  29 (require 'chise-subr)
  30 (require 'ideograph-subr)
  31
  32 (defconst unidata-normative-category-alist
  33   '(("Lu" letter        uppercase)
  34     ("Ll" letter        lowercase)
  35     ("Lt" letter        titlecase)
  36     ("Mn" mark          non-spacing)
  37     ("Mc" mark          spacing-combining)
  38     ("Me" mark          enclosing)
  39     ("Nd" number        decimal-digit)
  40     ("Nl" number        letter)
  41     ("No" number        other)
  42     ("Zs" separator     space)
  43     ("Zl" separator     line)
  44     ("Zp" separator     paragraph)
  45     ("Cc" other         control)
  46     ("Cf" other         format)
  47     ("Cs" other         surrogate)
  48     ("Co" other         private-use)
  49     ("Cn" other         not-assigned)))
  50
  51 (defconst unidata-informative-category-alist
  52   '(("Lm" letter        modifier)
  53     ("Lo" letter        other)
  54     ("Pc" punctuation   connector)
  55     ("Pd" punctuation   dash)
  56     ("Ps" punctuation   open)
  57     ("Pe" punctuation   close)
  58     ("Pi" punctuation   initial-quote)
  59     ("Pf" punctuation   final-quote)
  60     ("Po" punctuation   other)
  61     ("Sm" symbol        math)
  62     ("Sc" symbol        currency)
  63     ("Sk" symbol        modifier)
  64     ("So" symbol        other)
  65     ))
  66
  67 (defconst shuowen-radicals
  68   [?一 ?上 ?示 ?三 ?王 ?玉 ?玨 ?气 ?士 ?丨 ; 010
  69    ?屮 ?艸 ?蓐 ?茻 ?小 ?八 ?釆 ?半 ?牛 ?犛 ; 020
  70    ?告 ?口 ?凵 ?吅 ?哭 ?走 ?止 ?癶 ?步 ?此 ; 030
  71    ?正 ?是 ?辵 ?彳 ?廴 ?㢟 ?行 ?齒 ?牙 ?足 ; 040
  72    ?疋 ?品 ?龠 ?冊 ?㗊 ?舌 ?干 ?𧮫 ?只 ?㕯 ; 050
  73    ?句 ?丩 ?古 ?十 ?卅 ?言 ?誩 ?音 ?䇂 ?丵 ; 060
  74    ?菐 ?廾 ?𠬜 ?共 ?異 ?舁 ?𦥑 ?䢅 ?爨 ?革 ; 070
  75    ?鬲 ?䰜 ?爪 ?𠃨 ?鬥 ?又 ?𠂇 ?史 ?支 ?𦘒 ; 080
  76    ?聿 ?畫 ?隶 ?臤 ?臣 ?殳 ?殺 ?𠘧 ?寸 ?皮 ; 090
  77    ?㼱 ?攴 ?敎 ?卜 ?用 ?爻 ?㸚 ?𥄎 ?目 ?䀠 ; 100
  78    ?眉 ?盾 ?自 ?白 ?鼻 ?皕 ?習 ?羽 ?隹 ?奞 ; 110
  79    ?雈 ?𦫳 ?𥄕 ?羊 ?羴 ?瞿 ?雔 ?雥 ?鳥 ?烏 ; 120
  80    ?𠦒 ?冓 ?幺 ?𢆶 ?叀 ?玄 ?予 ?放 ?𠬪 ?𣦼 ; 130
  81    ?歺 ?死 ?冎 ?骨 ?肉 ?筋 ?刀 ?刃 ?㓞 ?丯 ; 140
  82    ?耒 ?角 ?竹 ?箕 ?丌 ?左 ?工 ?㠭 ?巫 ?甘 ; 150
  83    ?曰 ?乃 ?丂 ?可 ?兮 ?号 ?亏 ?旨 ?喜 ?壴 ; 160
  84 ;  ?旨 ?曰 ?乃 ?丂 ?可 ?兮 ?号 ?亏 ?喜 ?壴 ; 160
  85    ?鼓 ?豈 ?豆 ?豊 ?豐 ?䖒 ?虍 ?虎 ?虤 ?皿 ; 170
  86    ?𠙴 ?去 ?血 ?丶 ?丹 ?青 ?井 ?皀 ?鬯 ?食 ; 180
  87    ?亼 ?會 ?倉 ?入 ?缶 ?矢 ?高 ?冂 ?𩫏 ?京 ; 190
  88    ?亯 ?𣆪 ?畗 ?㐭 ?嗇 ?來 ?麥 ?夊 ?舛 ?䑞 ; 200
  89    ?韋 ?弟 ?夂 ?久 ?桀 ?木 ?東 ?林 ?才 ?叒 ; 210
  90    ?之 ?帀 ?出 ?𣎵 ?生 ?乇 ?𠂹 ?𠌶 ?華 ?𥝌 ; 220
  91    ?稽 ?巢 ?桼 ?束 ?㯻 ?囗 ?員 ?貝 ?邑 ?𨛜 ; 230
  92    ?日 ?旦 ?倝 ?㫃 ?冥 ?晶 ?月 ?有 ?明 ?囧 ; 240
  93    ?夕 ?多 ?毌 ?𢎘 ?𣐺 ?𠧪 ?齊 ?朿 ?片 ?鼎 ; 250
  94    ?克 ?彔 ?禾 ?秝 ?黍 ?香 ?米 ?毇 ?臼 ?凶 ; 260
  95    ?𣎳 ?𣏟 ?麻 ?尗 ?耑 ?韭 ?瓜 ?瓠 ?宀 ?宮 ; 270
  96    ?呂 ?穴 ?㝱 ?疒 ?冖 ?𠔼 ?冃 ?㒳 ?网 ?襾 ; 280
  97    ?巾 ?巿 ?帛 ?白 ?㡀 ?黹 ?人 ?𠤎 ?匕 ?从 ; 290
  98    ?比 ?北 ?丘 ?㐺 ?𡈼 ?重 ?臥 ?身 ?㐆 ?衣 ; 300
  99    ?裘 ?老 ?毛 ?毳 ?尸 ?尺 ?尾 ?履 ?舟 ?方 ; 310
 100    ?儿 ?兄 ?兂 ?皃 ?𠑹 ?先 ?秃 ?見 ?覞 ?欠 ; 320
 101    ?㱃 ?㳄 ?旡 ?頁 ?𦣻 ?面 ?丏 ?首 ?𥄉 ?須 ; 330
 102    ?彡 ?彣 ?文 ?髟 ?后 ?司 ?卮 ?卩 ?印 ?色 ; 340
 103    ?𠨍 ?辟 ?勹 ?包 ?茍 ?鬼 ?甶 ?厶 ?嵬 ?山 ; 350
 104    ?屾 ?屵 ?广 ?厂 ?丸 ?危 ?石 ?長 ?勿 ?冄 ; 360
 105    ?而 ?豕 ?㣇 ?彑 ?豚 ?豸 ?𤉡 ?易 ?象 ?馬 ; 370
 106    ?𢊁 ?鹿 ?麤 ?㲋 ?兔 ?萈 ?犬 ?㹜 ?鼠 ?能 ; 380
 107    ?熊 ?火 ?炎 ?黑 ?囪 ?焱 ?炙 ?赤 ?大 ?亦 ; 390
 108    ?夨 ?夭 ?交 ?尣 ?壺 ?壹 ?㚔 ?奢 ?亢 ?夲 ; 400
 109    ])
 110
 111 (defun shuowen-radical (number)
 112   (aref shuowen-radicals (1- number)))
 113
 114 (defvar char-db-file-coding-system 'utf-8-mcs-er)
 115
 116 (defvar char-db-ignored-attributes '(ideographic-products))
 117
 118 (defvar char-db-coded-charset-priority-list
 119   '(ascii
 120     control-1
 121     latin-iso8859-1
 122     latin-iso8859-2
 123     latin-iso8859-3
 124     latin-iso8859-4
 125     latin-iso8859-9
 126     latin-jisx0201
 127     cyrillic-iso8859-5
 128     greek-iso8859-7
 129     thai-tis620
 130     ;; =mj
 131     ;; =adobe-japan1-0
 132     ;; =adobe-japan1-1
 133     ;; =adobe-japan1-2
 134     ;; =adobe-japan1-3
 135     ;; =adobe-japan1-4
 136     ;; =adobe-japan1-5
 137     ;; =adobe-japan1-6
 138     =jis-x0208
 139     =jis-x0208@1978
 140     =jis-x0208@1983
 141     japanese-jisx0212
 142     chinese-gb2312
 143     =jis-x0208@1990
 144     chinese-cns11643-1
 145     chinese-cns11643-2
 146     chinese-cns11643-3
 147     chinese-cns11643-4
 148     chinese-cns11643-5
 149     chinese-cns11643-6
 150     chinese-cns11643-7
 151     =jis-x0213-1
 152     =jis-x0213-1@2000
 153     =jis-x0213-1@2004
 154     =jis-x0213-2
 155     korean-ksc5601
 156     chinese-isoir165
 157     katakana-jisx0201
 158     hebrew-iso8859-8
 159     chinese-gb12345
 160     latin-viscii
 161     ethiopic-ucs
 162     =big5-cdp
 163     =hanyo-denshi/ja
 164     =hanyo-denshi/jb
 165     =hanyo-denshi/jc
 166     =hanyo-denshi/jd
 167     =hanyo-denshi/ft
 168     =hanyo-denshi/ia
 169     =hanyo-denshi/ib
 170     =hanyo-denshi/hg
 171     =hanyo-denshi/jt
 172     =hanyo-denshi/ks
 173     =hanyo-denshi/tk
 174     ;; ==mj
 175     ;; ==adobe-japan1-0
 176     ;; ==adobe-japan1-1
 177     ;; ==adobe-japan1-2
 178     ;; ==adobe-japan1-3
 179     ;; ==adobe-japan1-4
 180     ;; ==adobe-japan1-5
 181     ;; ==adobe-japan1-6
 182     ==jis-x0208
 183     ==jis-x0213-1
 184     ==jis-x0213-2
 185     ==jis-x0212
 186     ==hanyo-denshi/ja
 187     ==hanyo-denshi/jb
 188     ==hanyo-denshi/jc
 189     ==hanyo-denshi/ft
 190     ==hanyo-denshi/ia
 191     ==hanyo-denshi/ib
 192     ==hanyo-denshi/hg
 193     ==hanyo-denshi/jt
 194     ==hanyo-denshi/ks
 195     ==hanyo-denshi/tk
 196     =gt
 197     =gt-k
 198     =daikanwa
 199     =daijiten
 200     =daikanwa@rev2
 201     =daikanwa@rev1
 202     =daikanwa/+p
 203     ==daikanwa
 204     ==daijiten
 205     =cbeta
 206     ideograph-hanziku-1
 207     ideograph-hanziku-2
 208     ideograph-hanziku-3
 209     ideograph-hanziku-4
 210     ideograph-hanziku-5
 211     ideograph-hanziku-6
 212     ideograph-hanziku-7
 213     ideograph-hanziku-8
 214     ideograph-hanziku-9
 215     ideograph-hanziku-10
 216     ideograph-hanziku-11
 217     ideograph-hanziku-12
 218     ;; =>>>adobe-japan1-0
 219     ;; =>>>adobe-japan1-1
 220     ;; =>>>adobe-japan1-2
 221     ;; =>>>adobe-japan1-3
 222     ;; =>>>adobe-japan1-4
 223     ;; =>>>adobe-japan1-5
 224     ;; =>>>adobe-japan1-6
 225     ;; =>>>jis-x0208
 226     ;; =>>>jis-x0213-1
 227     ;; =>>>jis-x0213-2
 228     ;; =>>>hanyo-denshi/ja
 229     ;; =>>>hanyo-denshi/jb
 230     ;; =>>>hanyo-denshi/jc
 231     ;; =>>>hanyo-denshi/ft
 232     ;; =>>>hanyo-denshi/ib
 233     ;; =>>>hanyo-denshi/hg
 234     ;; =>>>hanyo-denshi/jt
 235     ;; =>>>hanyo-denshi/ks
 236     ;; =>>>gt
 237     =>>adobe-japan1-0
 238     =>>adobe-japan1-1
 239     =>>adobe-japan1-2
 240     =>>adobe-japan1-3
 241     =>>adobe-japan1-4
 242     =>>adobe-japan1-5
 243     =>>adobe-japan1-6
 244     =>>jis-x0208
 245     =>>jis-x0213-1
 246     =>>jis-x0213-1@2000
 247     =>>jis-x0213-1@2004
 248     =>>jis-x0213-2
 249     =>>jis-x0208@1978
 250     =>>hanyo-denshi/ft
 251     =>>hanyo-denshi/jt
 252     =>>hanyo-denshi/ks
 253     =>>gt
 254     =>>daikanwa
 255     =>>cbeta
 256     =+>jis-x0208
 257     =+>jis-x0213-1
 258     =+>jis-x0213-2
 259     =+>adobe-japan1-0
 260     =+>adobe-japan1-1
 261     =+>adobe-japan1-2
 262     =+>adobe-japan1-3
 263     =+>adobe-japan1-4
 264     =+>adobe-japan1-5
 265     =+>adobe-japan1-6
 266     =+>jis-x0208@1978
 267     =>jis-x0208
 268     =>jis-x0208@1997
 269     =>jis-x0213-1
 270     =>jis-x0213-1@2000
 271     =>jis-x0213-1@2004
 272     =>jis-x0213-2
 273     ==>ucs@bucs
 274     ==>daijiten
 275     ;; =>ucs@hanyo-denshi
 276     =>ucs@iso
 277     =>ucs@unicode
 278     =>ucs@jis
 279     =>ucs@cns
 280     =>ucs@ks
 281     =+>ucs@iso
 282     =+>ucs@unicode
 283     =+>ucs@jis
 284     =+>ucs@jis/1990
 285     =+>ucs@cns
 286     =+>ucs@ks
 287     =>>ucs@iso
 288     =>>ucs@unicode
 289     =>>ucs@jis
 290     =>>ucs@cns
 291     =>>>ucs@iso
 292     =>>>ucs@unicode
 293     ==ucs@iso
 294     ==ucs@unicode
 295     ;; ==ucs@cns
 296     ==gb2312
 297     ==ks-x1001
 298     ==cns11643-1
 299     ==cns11643-2
 300     ==cns11643-3
 301     ==cns11643-4
 302     ==cns11643-5
 303     ==cns11643-6
 304     ==cns11643-7
 305     ==gt
 306     ==jis-x0208@1990
 307     ;; ==jis-x0208@1983
 308     ==jis-x0208@1978
 309     ==gt-k
 310     =ucs@iso
 311     =ucs@unicode
 312     =ucs@cns
 313     ==big5-cdp
 314     ==cbeta
 315     =>>big5-cdp
 316     =>>gt-k
 317     =+>gt
 318     =+>big5-cdp
 319     =>big5-cdp
 320     =>big5-cdp@iwds-1
 321     =>mj
 322     =>iwds-1
 323     =big5
 324     =big5-eten
 325     =zinbun-oracle
 326     =>zinbun-oracle
 327     =ruimoku-v6
 328     =>>ruimoku-v6
 329     ==ruimoku-v6
 330     =jef-china3
 331     =shinjigen
 332     =>cns11643-5
 333     =>cns11643-7
 334     =>gt
 335     =>gt-k
 336     =>daikanwa
 337     =>daikanwa/ho
 338     =>cbeta
 339     =ucs-var-001
 340     =ucs-var-002
 341     =ucs-var-003
 342     =ucs-var-004
 343     =ucs-var-005
 344     =ucs-var-006
 345     =ucs-var-007
 346     =ucs-var-008
 347     =ucs-var-010
 348     =ucs-itaiji-001
 349     =ucs-itaiji-002
 350     =ucs-itaiji-003
 351     =ucs-itaiji-004
 352     =ucs-itaiji-005
 353     =ucs-itaiji-006
 354     =ucs-itaiji-007
 355     =ucs-itaiji-008
 356     =ucs-itaiji-009
 357     =ucs-itaiji-010
 358     =ucs-itaiji-011
 359     =ucs-itaiji-012
 360     =+>ucs-var-003
 361     =>ucs-itaiji-001
 362     =>ucs-itaiji-002
 363     =>ucs-itaiji-003
 364     =>ucs-itaiji-004
 365     =>ucs-itaiji-005
 366     =>ucs-itaiji-006
 367     =>ucs-itaiji-007
 368     =>ucs-itaiji-008
 369     =>ucs-itaiji-009
 370     =>ucs-itaiji-010
 371     =>ucs-itaiji-011
 372     =big5-cdp-var-001
 373     =big5-cdp-var-002
 374     =big5-cdp-var-003
 375     =big5-cdp-var-004
 376     =big5-cdp-var-005
 377     =big5-cdp-var-007
 378     =big5-cdp-var-010
 379     =big5-cdp-itaiji-001
 380     =big5-cdp-itaiji-002
 381     =big5-cdp-itaiji-003
 382     =>big5-cdp-itaiji-001
 383     =>big5-cdp-itaiji-003
 384     =>big5-cdp-var-001
 385     =>ucs@iwds-1
 386     =>ucs@cognate
 387     =>ucs@component
 388     =>ucs@iwds-1/normalized
 389     =>ucs-itaiji-001@iwds-1
 390     =>ucs-itaiji-002@iwds-1
 391     =>big5-cdp@component
 392     =>big5-cdp@cognate
 393     ==ucs@gb
 394     =ucs@gb
 395     ==ucs-var-002
 396     =ucs@JP/hanazono))
 397
 398
 399 ;;; @ char-db formatters
 400 ;;;
 401
 402 (defun char-db-make-char-spec (char)
 403   (let (ret char-spec)
 404     (cond ((characterp char)
 405            (cond ((and (setq ret (encode-char char '=ucs 'defined-only))
 406                        (not (and (<= #xE000 ret)(<= ret #xF8FF))))
 407                   (setq char-spec (list (cons '=ucs ret)))
 408                   (cond ((setq ret (get-char-attribute char 'name))
 409                          (setq char-spec (cons (cons 'name ret) char-spec))
 410                          )
 411                         ((setq ret (get-char-attribute char 'name*))
 412                          (setq char-spec (cons (cons 'name* ret) char-spec))
 413                          ))
 414                   )
 415                  ((encode-char char '=mj 'defined-only)
 416                   (setq char-spec nil)
 417                   (dolist (ccs (charset-list))
 418                     (if (and (or (eq ccs '=mj)
 419                                  ;; (eq (charset-property ccs 'iso-ir) 177)
 420                                  (string-match "=ucs@" (symbol-name ccs))
 421                                  )
 422                              (setq ccs (charset-name ccs))
 423                              (null (assq ccs char-spec))
 424                              (setq ret (encode-char char ccs 'defined-only)))
 425                         (setq char-spec (cons (cons ccs ret) char-spec))))
 426                   )
 427                  ((encode-char char '==mj 'defined-only)
 428                   (setq char-spec nil)
 429                   (dolist (ccs (charset-list))
 430                     (if (and (or (eq ccs '==mj)
 431                                  ;; (eq (charset-property ccs 'iso-ir) 177)
 432                                  (string-match "=ucs@" (symbol-name ccs))
 433                                  )
 434                              (setq ccs (charset-name ccs))
 435                              (null (assq ccs char-spec))
 436                              (setq ret (encode-char char ccs 'defined-only)))
 437                         (setq char-spec (cons (cons ccs ret) char-spec))))
 438                   )
 439                  ((encode-char char '=adobe-japan1 'defined-only)
 440                   (setq char-spec nil)
 441                   (dolist (ccs (charset-list))
 442                     (if (and (or (memq ccs
 443                                        '(=adobe-japan1-0
 444                                          =adobe-japan1-1
 445                                          =adobe-japan1-2
 446                                          =adobe-japan1-3
 447                                          =adobe-japan1-4
 448                                          =adobe-japan1-5
 449                                          =adobe-japan1-6
 450                                          ))
 451                                  ;; (eq (charset-property ccs 'iso-ir) 177)
 452                                  (string-match "=ucs@" (symbol-name ccs))
 453                                  )
 454                              (setq ccs (charset-name ccs))
 455                              (null (assq ccs char-spec))
 456                              (setq ret (encode-char char ccs 'defined-only)))
 457                         (setq char-spec (cons (cons ccs ret) char-spec))))
 458                   )
 459                  ((encode-char char '==adobe-japan1 'defined-only)
 460                   (setq char-spec nil)
 461                   (dolist (ccs (charset-list))
 462                     (if (and (or (memq ccs
 463                                        '(==adobe-japan1-0
 464                                          ==adobe-japan1-1
 465                                          ==adobe-japan1-2
 466                                          ==adobe-japan1-3
 467                                          ==adobe-japan1-4
 468                                          ==adobe-japan1-5
 469                                          ==adobe-japan1-6
 470                                          ))
 471                                  ;; (eq (charset-property ccs 'iso-ir) 177)
 472                                  (string-match "=ucs@" (symbol-name ccs))
 473                                  )
 474                              (setq ccs (charset-name ccs))
 475                              (null (assq ccs char-spec))
 476                              (setq ret (encode-char char ccs 'defined-only)))
 477                         (setq char-spec (cons (cons ccs ret) char-spec))))
 478                   )
 479                  ((setq ret
 480                         (catch 'tag
 481                           (let ((rest char-db-coded-charset-priority-list)
 482                                 ccs)
 483                             (while rest
 484                               (setq ccs (charset-name
 485                                          (find-charset (car rest))))
 486                               (if (setq ret
 487                                         (encode-char char ccs
 488                                                      'defined-only))
 489                                   (throw 'tag (cons ccs ret)))
 490                               (setq rest (cdr rest))))))
 491                   (setq char-spec (list ret))
 492                   (dolist (ccs (delq (car ret) (charset-list)))
 493                     (if (and (or (charset-iso-final-char ccs)
 494                                  (memq ccs
 495                                        '(=daikanwa
 496                                          =daikanwa@rev2
 497                                          ;; =gt-k
 498                                          =jis-x0208@1997
 499                                          ))
 500                                  (and (eq (charset-property ccs 'iso-ir) 177)
 501                                       (not (eq (charset-name ccs) '=ucs@big5)))
 502                                  ;; (string-match "=ucs@" (symbol-name ccs))
 503                                  )
 504                              (setq ccs (charset-name ccs))
 505                              (null (assq ccs char-spec))
 506                              (setq ret (encode-char char ccs 'defined-only)))
 507                         (setq char-spec (cons (cons ccs ret) char-spec))))
 508                   (if (null char-spec)
 509                       (setq char-spec (split-char char)))
 510                   (cond ((setq ret (get-char-attribute char 'name))
 511                          (setq char-spec (cons (cons 'name ret) char-spec))
 512                          )
 513                         ((setq ret (get-char-attribute char 'name*))
 514                          (setq char-spec (cons (cons 'name* ret) char-spec))
 515                          ))
 516                   )
 517                  ((setq ret (get-char-attribute
 518                              char 'ideographic-combination))
 519                   (setq char-spec
 520                         (cons (cons 'ideographic-combination ret)
 521                               char-spec))
 522                   ))
 523            char-spec)
 524           ((consp char)
 525            char))))
 526
 527 (defun char-db-insert-char-spec (char &optional readable column
 528                                       required-features)
 529   (unless column
 530     (setq column (current-column)))
 531   (let (char-spec temp-char)
 532     (setq char-spec (char-db-make-char-spec char))
 533     (unless (or (characterp char) ; char
 534                 (condition-case nil
 535                     (setq char (find-char char-spec))
 536                   (error nil)))
 537       ;; define temporary character
 538       ;;   Current implementation is dirty.
 539       (setq temp-char (define-char (cons '(ideograph-daikanwa . 0)
 540                                          char-spec)))
 541       (remove-char-attribute temp-char 'ideograph-daikanwa)
 542       (setq char temp-char))
 543     (insert-char-attributes char
 544                             readable
 545                             (union (mapcar #'car char-spec)
 546                                    required-features)
 547                             nil 'for-sub-node)
 548     (when temp-char
 549       ;; undefine temporary character
 550       ;;   Current implementation is dirty.
 551       (setq char-spec (char-attribute-alist temp-char))
 552       (while char-spec
 553         (remove-char-attribute temp-char (car (car char-spec)))
 554         (setq char-spec (cdr char-spec))))))
 555
 556 (defun char-db-insert-alist (alist &optional readable column)
 557   (unless column
 558     (setq column (current-column)))
 559   (let ((line-breaking
 560          (concat "\n" (make-string (1+ column) ?\ )))
 561         name value
 562         ret al ; cal
 563         key
 564         lbs cell rest separator)
 565     (insert "(")
 566     (while alist
 567       (setq name (car (car alist))
 568             value (cdr (car alist)))
 569       (cond ((eq name 'char)
 570              (insert "(char . ")
 571              (if (setq ret (condition-case nil
 572                                (find-char value)
 573                              (error nil)))
 574                  (progn
 575                    (setq al nil
 576                          ;; cal nil
 577                          )
 578                    (while value
 579                      (setq key (car (car value)))
 580                      ;; (if (find-charset key)
 581                      ;;     (setq cal (cons key cal))
 582                      (setq al (cons key al))
 583                      ;; )
 584                      (setq value (cdr value)))
 585                    (insert-char-attributes ret
 586                                            readable
 587                                            (or al 'none) ; cal
 588                                            nil 'for-sub-node))
 589                (insert (prin1-to-string value)))
 590              (insert ")")
 591              (insert line-breaking))
 592             ((consp value)
 593              (insert (format "(%-18s " name))
 594              (setq lbs (concat "\n" (make-string (current-column) ?\ )))
 595              (while (consp value)
 596                (setq cell (car value))
 597                (if (and (consp cell)
 598                         (consp (car cell))
 599                         (setq ret (condition-case nil
 600                                       (find-char cell)
 601                                     (error nil)))
 602                         )
 603                    (progn
 604                      (setq rest cell
 605                            al nil
 606                            ;; cal nil
 607                            )
 608                      (while rest
 609                        (setq key (car (car rest)))
 610                        ;; (if (find-charset key)
 611                        ;;     (setq cal (cons key cal))
 612                        (setq al (cons key al))
 613                        ;; )
 614                        (setq rest (cdr rest)))
 615                      (if separator
 616                          (insert lbs))
 617                      (insert-char-attributes ret
 618                                              readable
 619                                              al ; cal
 620                                              nil 'for-sub-node)
 621                      (setq separator lbs))
 622                  (if separator
 623                      (insert separator))
 624                  (insert (prin1-to-string cell))
 625                  (setq separator " "))
 626                (setq value (cdr value)))
 627              (insert ")")
 628              (insert line-breaking))
 629             (t
 630              (insert (format "(%-18s . %S)%s"
 631                              name value
 632                              line-breaking))))
 633       (setq alist (cdr alist))))
 634   (insert ")"))
 635
 636 (defun char-db-insert-char-reference (plist &optional readable column)
 637   (unless column
 638     (setq column (current-column)))
 639   (let ((line-breaking
 640          (concat "\n" (make-string (1+ column) ?\ )))
 641         (separator "")
 642         name value)
 643     (insert "(")
 644     (while plist
 645       (setq name (pop plist))
 646       (setq value (pop plist))
 647       (cond ((eq name :char)
 648              (insert separator)
 649              (insert ":char\t")
 650              (cond ((numberp value)
 651                     (setq value (decode-char '=ucs value)))
 652                    ;; ((consp value)
 653                    ;;  (setq value (or (find-char value)
 654                    ;;                  value)))
 655                    )
 656              (char-db-insert-char-spec value readable)
 657              (insert line-breaking)
 658              (setq separator ""))
 659             ((eq name :radical)
 660              (insert (format "%s%s\t%d ; %c%s"
 661                              separator
 662                              name value
 663                              (ideographic-radical value)
 664                              line-breaking))
 665              (setq separator ""))
 666             (t
 667              (insert (format "%s%s\t%S" separator name value))
 668              (setq separator line-breaking)))
 669       ))
 670   (insert ")"))
 671
 672 (defun char-db-decode-isolated-char (ccs code-point)
 673   (let (ret)
 674     (setq ret
 675           (cond ((eq ccs 'arabic-iso8859-6)
 676                  (decode-char ccs code-point))
 677                 ;; ((eq ccs '=gt)
 678                 ;;  (decode-builtin-char '==gt code-point))
 679                 ((and (memq ccs '(=gt-pj-1
 680                                   =gt-pj-2
 681                                   =gt-pj-3
 682                                   =gt-pj-4
 683                                   =gt-pj-5
 684                                   =gt-pj-6
 685                                   =gt-pj-7
 686                                   =gt-pj-8
 687                                   =gt-pj-9
 688                                   =gt-pj-10
 689                                   =gt-pj-11))
 690                       (setq ret (decode-char ccs code-point))
 691                       (setq ret (encode-char ret '=gt 'defined-only)))
 692                  (decode-builtin-char '=gt ret))
 693                 (t
 694                  (decode-builtin-char ccs code-point))))
 695     (cond ((null ret)
 696            (or (decode-char ccs code-point)
 697                (define-char (list (cons ccs code-point)))))
 698           ((and (<= 0 (char-int ret))
 699                 (<= (char-int ret) #x1F))
 700            (decode-char '=ucs (+ #x2400 (char-int ret))))
 701           ((= (char-int ret) #x7F)
 702            ?\u2421)
 703           (t ret))))
 704
 705 (defvar char-db-convert-obsolete-format t)
 706
 707 (defun char-db-insert-ccs-feature (name value line-breaking)
 708   (cond
 709    ((integerp value)
 710     (insert
 711      (format
 712       (cond
 713        ((memq name '(=>iwds-1
 714                      ==shinjigen
 715                      =shinjigen
 716                      =shinjigen@1ed ==shinjigen@1ed
 717                      =shinjigen@rev ==shinjigen@rev
 718                      =shinjigen/+p@rev ==shinjigen/+p@rev
 719                      ===daikanwa/ho ==daikanwa/ho
 720                      =daikanwa/ho =>>daikanwa/ho =>daikanwa/ho))
 721         "(%-18s .  %04d)\t; %c")
 722        ((eq name '=shinjigen@1ed/24pr)
 723         "(%-18s . %04d)\t; %c")
 724        ((or
 725          (memq name
 726                '(===daikanwa
 727                  ==daikanwa =daikanwa =>>daikanwa =>daikanwa
 728                  =daikanwa@rev1 =daikanwa@rev2
 729                  =daikanwa/+p ==daikanwa/+p ===daikanwa/+p
 730                  =>>daikanwa/+p
 731                  =daikanwa/+2p =>>daikanwa/+2p
 732                  =gt ==gt ===gt
 733                  =>>gt =+>gt =>gt
 734                  =gt-k ==gt-k ===gt-k
 735                  =>>gt-k =>gt-k
 736                  =adobe-japan1-0 ==adobe-japan1-0 ===adobe-japan1-0
 737                  =adobe-japan1-1 ==adobe-japan1-1 ===adobe-japan1-1
 738                  =adobe-japan1-2 ==adobe-japan1-2 ===adobe-japan1-2
 739                  =adobe-japan1-3 ==adobe-japan1-3 ===adobe-japan1-3
 740                  =adobe-japan1-4 ==adobe-japan1-4 ===adobe-japan1-4
 741                  =adobe-japan1-5 ==adobe-japan1-5 ===adobe-japan1-5
 742                  =adobe-japan1-6 ==adobe-japan1-6 ===adobe-japan1-6
 743                  =>>adobe-japan1-0 =+>adobe-japan1-0
 744                  =>>adobe-japan1-1 =+>adobe-japan1-1
 745                  =>>adobe-japan1-2 =+>adobe-japan1-2
 746                  =>>adobe-japan1-3 =+>adobe-japan1-3
 747                  =>>adobe-japan1-4 =+>adobe-japan1-4
 748                  =>>adobe-japan1-5 =+>adobe-japan1-5
 749                  =>>adobe-japan1-6 =+>adobe-japan1-6
 750                  =>cbeta =cbeta =>>cbeta ==cbeta ===cbeta
 751                  =zinbun-oracle =>zinbun-oracle
 752                  =daijiten ==daijiten ===daijiten ==>daijiten
 753                  ===hng-jou ===hng-keg ===hng-dng ===hng-mam
 754                  ===hng-drt ===hng-kgk ===hng-myz ===hng-kda
 755                  ===hng-khi ===hng-khm ===hng-fhs ===hng-hok
 756                  ===hng-kyd ===hng-sok
 757                  ===hng-yhk ===hng-kak ===hng-kar ===hng-kae
 758                  ===hng-sys ===hng-tsu ===hng-tzj
 759                  ===hng-hos ===hng-kkh ===hng-nak ===hng-jhk
 760                  ===hng-hod ===hng-gok ===hng-ink ===hng-nto
 761                  ===hng-nkm ===hng-k24 ===hng-ini ===hng-nkk
 762                  ===hng-kcc ===hng-kcj ===hng-kbk ===hng-sik
 763                  ===hng-skk ===hng-kyu ===hng-ksk ===hng-wan
 764                  ===hng-okd ===hng-wad ===hng-kmi
 765                  ===hng-zkd ===hng-doh ===hng-jyu ===hng-tzs
 766                  ===hng-sai ===hng-kad ===hng-kss
 767                  ===hng-kyo ===hng-ykk ===hng-saa
 768                  ===hng-sab ===hng-wks ===hng-wke ===hng-smk
 769                  ===hng-sgs ===hng-sts
 770                  =shuowen-jiguge ===shuowen-jiguge4 ===shuowen-jiguge5))
 771          ;; (string-match "^=adobe-" (symbol-name name))
 772          )
 773         "(%-18s . %05d)\t; %c")
 774        ((memq name '(=hanyo-denshi/ks
 775                      ==hanyo-denshi/ks ===hanyo-denshi/ks
 776                      =>>hanyo-denshi/ks
 777                      =koseki ==koseki
 778                      =mj ==mj ===mj =>>mj =>mj
 779                      =zihai mojikyo))
 780         "(%-18s . %06d)\t; %c")
 781        ((memq name '(=hanyo-denshi/tk ==hanyo-denshi/tk))
 782         "(%-18s . %08d)\t; %c")
 783        ((>= (charset-dimension name) 2)
 784         "(%-18s . #x%04X)\t; %c")
 785        (t
 786         "(%-18s . #x%02X)\t; %c"))
 787       name
 788       (if (= (charset-iso-graphic-plane name) 1)
 789           (logior value
 790                   (cond ((= (charset-dimension name) 1)
 791                          #x80)
 792                         ((= (charset-dimension name) 2)
 793                          #x8080)
 794                         ((= (charset-dimension name) 3)
 795                          #x808080)
 796                         (t 0)))
 797         value)
 798       (char-db-decode-isolated-char name value)))
 799     (if (and (= (charset-chars name) 94)
 800              (= (charset-dimension name) 2))
 801         (insert (format " [%02d-%02d]"
 802                         (- (lsh value -8) 32)
 803                         (- (logand value 255) 32))))
 804     )
 805    (t
 806     (insert (format "(%-18s . %s)" name value))
 807     ))
 808   (insert line-breaking))
 809
 810 (defun char-db-insert-relation-feature (char name value line-breaking
 811                                              ccss readable)
 812   (insert (format "(%-18s%s " name line-breaking))
 813   (let ((lbs (concat "\n" (make-string (current-column) ?\ )))
 814         separator cell sources required-features
 815         ret)
 816     (while (consp value)
 817       (setq cell (car value))
 818       (if (integerp cell)
 819           (setq cell (decode-char '=ucs cell)))
 820       (cond
 821        ((eq name '->subsumptive)
 822         (when (or (not (some (lambda (atr)
 823                                (get-char-attribute cell atr))
 824                              char-db-ignored-attributes))
 825                   (some (lambda (ccs)
 826                           (encode-char cell ccs 'defined-only))
 827                         ccss))
 828           (if separator
 829               (insert lbs))
 830           (let ((char-db-ignored-attributes
 831                  (cons '<-subsumptive
 832                        char-db-ignored-attributes)))
 833             (insert-char-attributes cell readable nil nil 'for-sub-node))
 834           (setq separator lbs))
 835         )
 836        ((characterp cell)
 837         (setq sources
 838               (get-char-attribute
 839                char (intern (format "%s*sources" name))))
 840         (setq required-features nil)
 841         (dolist (source sources)
 842           (cond
 843            ((memq source '(JP
 844                            JP/Jouyou
 845                            shinjigen shinjigen@1ed shinjigen@rev))
 846             (setq required-features
 847                   (union required-features
 848                          '(=jis-x0208
 849                            =jis-x0208@1990
 850                            =jis-x0213-1@2000
 851                            =jis-x0213-1@2004
 852                            =jis-x0213-2
 853                            =jis-x0212
 854                            =jis-x0208@1983
 855                            =jis-x0208@1978
 856                            =shinjigen
 857                            =shinjigen@1ed
 858                            =shinjigen@rev
 859                            =shinjigen/+p@rev))))
 860            ((eq source 'CN)
 861             (setq required-features
 862                   (union required-features
 863                          '(=gb2312
 864                            =gb12345
 865                            =iso-ir165)))))
 866           (cond
 867            ((find-charset (setq ret (intern (format "=%s" source))))
 868             (setq required-features
 869                   (cons ret required-features)))
 870            (t (setq required-features
 871                     (cons source required-features)))))
 872         (cond ((string-match "@JP" (symbol-name name))
 873                (setq required-features
 874                      (union required-features
 875                             '(=jis-x0208
 876                               =jis-x0208@1990
 877                               =jis-x0213-1-2000
 878                               =jis-x0213-2-2000
 879                               =jis-x0212
 880                               =jis-x0208@1983
 881                               =jis-x0208@1978))))
 882               ((string-match "@CN" (symbol-name name))
 883                (setq required-features
 884                      (union required-features
 885                             '(=gb2312
 886                               =gb12345
 887                               =iso-ir165)))))
 888         (if separator
 889             (insert lbs))
 890         (if readable
 891             (insert (format "%S" cell))
 892           (char-db-insert-char-spec cell readable
 893                                     nil
 894                                     required-features))
 895         (setq separator lbs))
 896        ((consp cell)
 897         (if separator
 898             (insert lbs))
 899         (if (consp (car cell))
 900             (char-db-insert-char-spec cell readable)
 901           (char-db-insert-char-reference cell readable))
 902         (setq separator lbs))
 903        (t
 904         (if separator
 905             (insert separator))
 906         (insert (prin1-to-string cell))
 907         (setq separator " ")))
 908       (setq value (cdr value)))
 909     (insert ")")
 910     (insert line-breaking)))
 911
 912 (defun insert-char-attributes (char &optional readable attributes column
 913                                     for-sub-node)
 914   (unless column
 915     (setq column (current-column)))
 916   (let (name value ; has-long-ccs-name
 917         rest
 918         radical strokes
 919         (line-breaking
 920          (concat "\n" (make-string (1+ column) ?\ )))
 921         lbs cell separator ret
 922         key al cal
 923         dest-ccss ; sources required-features
 924         ccss)
 925     (let (atr-d)
 926       (setq attributes
 927             (sort (if attributes
 928                       (if (consp attributes)
 929                           (progn
 930                             (dolist (name attributes)
 931                               (unless (memq name char-db-ignored-attributes)
 932                                 (if (find-charset name)
 933                                     (push name ccss))
 934                                 (push name atr-d)))
 935                             atr-d))
 936                     (dolist (name (char-attribute-list))
 937                       (unless (memq name char-db-ignored-attributes)
 938                         (if (find-charset name)
 939                             (push name ccss))
 940                         (push name atr-d)))
 941                     atr-d)
 942                   #'char-attribute-name<)))
 943     (insert "(")
 944     (when (memq '<-subsumptive attributes)
 945       (when (or readable (not for-sub-node))
 946         (when (setq value (get-char-attribute char '<-subsumptive))
 947           (char-db-insert-relation-feature char '<-subsumptive value
 948                                            line-breaking
 949                                            ccss readable)))
 950       (setq attributes (delq '<-subsumptive attributes)))
 951     (when (and (memq '<-denotational attributes)
 952                (setq value (get-char-attribute char '<-denotational)))
 953       (char-db-insert-relation-feature char '<-denotational value
 954                                        line-breaking
 955                                        ccss readable)
 956       (setq attributes (delq '<-denotational attributes)))
 957     (when (and (memq '<-denotational@component attributes)
 958                (setq value (get-char-attribute char '<-denotational@component)))
 959       (char-db-insert-relation-feature char '<-denotational@component value
 960                                        line-breaking
 961                                        ccss readable)
 962       (setq attributes (delq '<-denotational@component attributes)))
 963     (when (and (memq '<-denotational@usage attributes)
 964                (setq value (get-char-attribute char '<-denotational@usage)))
 965       (char-db-insert-relation-feature char '<-denotational@usage value
 966                                        line-breaking
 967                                        ccss readable)
 968       (setq attributes (delq '<-denotational@usage attributes)))
 969     (when (and (memq 'name attributes)
 970                (setq value (get-char-attribute char 'name)))
 971       (insert (format
 972                (if (> (+ (current-column) (length value)) 48)
 973                    "(name . %S)%s"
 974                  "(name               . %S)%s")
 975                value line-breaking))
 976       (setq attributes (delq 'name attributes))
 977       )
 978     (when (and (memq 'name* attributes)
 979                (setq value (get-char-attribute char 'name*)))
 980       (insert (format
 981                (if (> (+ (current-column) (length value)) 48)
 982                    "(name* . %S)%s"
 983                  "(name*              . %S)%s")
 984                value line-breaking))
 985       (setq attributes (delq 'name* attributes))
 986       )
 987     (when (and (memq 'script attributes)
 988                (setq value (get-char-attribute char 'script)))
 989       (insert (format "(script\t\t%s)%s"
 990                       (mapconcat (function prin1-to-string)
 991                                  value " ")
 992                       line-breaking))
 993       (setq attributes (delq 'script attributes))
 994       )
 995     (dolist (name '(=>ucs =>ucs*))
 996       (when (and (memq name attributes)
 997                  (setq value (get-char-attribute char name)))
 998         (insert (format "(%-18s . #x%04X)\t; %c%s"
 999                         name value (decode-char '=ucs value)
1000                         line-breaking))
1001         (setq attributes (delq name attributes))))
1002     (dolist (name '(=>ucs@gb =>ucs@big5))
1003       (when (and (memq name attributes)
1004                  (setq value (get-char-attribute char name)))
1005         (insert (format "(%-18s . #x%04X)\t; %c%s"
1006                         name value
1007                         (decode-char (intern
1008                                       (concat "="
1009                                               (substring
1010                                                (symbol-name name) 2)))
1011                                      value)
1012                         line-breaking))
1013         (setq attributes (delq name attributes))
1014         ))
1015     ;; (dolist (name '(=>daikanwa))
1016     ;;   (when (and (memq name attributes)
1017     ;;              (setq value (get-char-attribute char name)))
1018     ;;     (insert
1019     ;;      (if (integerp value)
1020     ;;          (format "(%-18s . %05d)\t; %c%s"
1021     ;;                  name value (decode-char '=daikanwa value)
1022     ;;                  line-breaking)
1023     ;;        (format "(%-18s %s)\t; %c%s"
1024     ;;                name
1025     ;;                (mapconcat (function prin1-to-string)
1026     ;;                           value " ")
1027     ;;                (char-representative-of-daikanwa char)
1028     ;;                line-breaking)))
1029     ;;     (setq attributes (delq name attributes))))
1030     (when (and (memq 'general-category attributes)
1031                (setq value (get-char-attribute char 'general-category)))
1032       (insert (format
1033                "(general-category\t%s) ; %s%s"
1034                (mapconcat (lambda (cell)
1035                             (format "%S" cell))
1036                           value " ")
1037                (cond ((rassoc value unidata-normative-category-alist)
1038                       "Normative Category")
1039                      ((rassoc value unidata-informative-category-alist)
1040                       "Informative Category")
1041                      (t
1042                       "Unknown Category"))
1043                line-breaking))
1044       (setq attributes (delq 'general-category attributes))
1045       )
1046     (when (and (memq 'bidi-category attributes)
1047                (setq value (get-char-attribute char 'bidi-category)))
1048       (insert (format "(bidi-category\t. %S)%s"
1049                       value
1050                       line-breaking))
1051       (setq attributes (delq 'bidi-category attributes))
1052       )
1053     (unless (or (not (memq 'mirrored attributes))
1054                 (eq (setq value (get-char-attribute char 'mirrored 'empty))
1055                     'empty))
1056       (insert (format "(mirrored\t\t. %S)%s"
1057                       value
1058                       line-breaking))
1059       (setq attributes (delq 'mirrored attributes))
1060       )
1061     (cond
1062      ((and (memq 'decimal-digit-value attributes)
1063            (setq value (get-char-attribute char 'decimal-digit-value)))
1064       (insert (format "(decimal-digit-value . %S)%s"
1065                       value
1066                       line-breaking))
1067       (setq attributes (delq 'decimal-digit-value attributes))
1068       (when (and (memq 'digit-value attributes)
1069                  (setq value (get-char-attribute char 'digit-value)))
1070         (insert (format "(digit-value\t . %S)%s"
1071                         value
1072                         line-breaking))
1073         (setq attributes (delq 'digit-value attributes))
1074         )
1075       (when (and (memq 'numeric-value attributes)
1076                  (setq value (get-char-attribute char 'numeric-value)))
1077         (insert (format "(numeric-value\t . %S)%s"
1078                         value
1079                         line-breaking))
1080         (setq attributes (delq 'numeric-value attributes))
1081         )
1082       )
1083      (t
1084       (when (and (memq 'digit-value attributes)
1085                  (setq value (get-char-attribute char 'digit-value)))
1086         (insert (format "(digit-value\t. %S)%s"
1087                         value
1088                         line-breaking))
1089         (setq attributes (delq 'digit-value attributes))
1090         )
1091       (when (and (memq 'numeric-value attributes)
1092                  (setq value (get-char-attribute char 'numeric-value)))
1093         (insert (format "(numeric-value\t. %S)%s"
1094                         value
1095                         line-breaking))
1096         (setq attributes (delq 'numeric-value attributes))
1097         )))
1098     (when (and (memq 'iso-10646-comment attributes)
1099                (setq value (get-char-attribute char 'iso-10646-comment)))
1100       (insert (format "(iso-10646-comment\t. %S)%s"
1101                       value
1102                       line-breaking))
1103       (setq attributes (delq 'iso-10646-comment attributes))
1104       )
1105     (when (and (memq 'morohashi-daikanwa attributes)
1106                (setq value (get-char-attribute char 'morohashi-daikanwa)))
1107       (insert (format "(morohashi-daikanwa\t%s)%s"
1108                       (mapconcat (function prin1-to-string) value " ")
1109                       line-breaking))
1110       (setq attributes (delq 'morohashi-daikanwa attributes))
1111       )
1112     (setq radical nil
1113           strokes nil)
1114     (when (and (memq 'ideographic-radical attributes)
1115                (setq value (get-char-attribute char 'ideographic-radical)))
1116       (setq radical value)
1117       (insert (format "(ideographic-radical . %S)\t; %c%s"
1118                       radical
1119                       (ideographic-radical radical)
1120                       line-breaking))
1121       (setq attributes (delq 'ideographic-radical attributes))
1122       )
1123     (when (and (memq 'shuowen-radical attributes)
1124                (setq value (get-char-attribute char 'shuowen-radical)))
1125       (insert (format "(shuowen-radical\t. %S)\t; %c%s"
1126                       value
1127                       (shuowen-radical value)
1128                       line-breaking))
1129       (setq attributes (delq 'shuowen-radical attributes))
1130       )
1131     (let (key)
1132       (dolist (domain
1133                (append
1134                 char-db-feature-domains
1135                 (let (dest domain)
1136                   (dolist (feature (char-attribute-list))
1137                     (setq feature (symbol-name feature))
1138                     (when (string-match
1139                            "\\(radical\\|strokes\\)@\\([^@*]+\\)\\(\\*\\|$\\)"
1140                            feature)
1141                       (setq domain (intern (match-string 2 feature)))
1142                      (unless (memq domain dest)
1143                        (setq dest (cons domain dest)))))
1144                   (sort dest #'string<))))
1145         (setq key (intern (format "%s@%s" 'ideographic-radical domain)))
1146         (when (and (memq key attributes)
1147                    (setq value (get-char-attribute char key)))
1148           (setq radical value)
1149           (insert (format "(%s . %S)\t; %c%s"
1150                           key
1151                           radical
1152                           (ideographic-radical radical)
1153                           line-breaking))
1154           (setq attributes (delq key attributes))
1155           )
1156         (setq key (intern (format "%s@%s" 'ideographic-strokes domain)))
1157         (when (and (memq key attributes)
1158                    (setq value (get-char-attribute char key)))
1159           (setq strokes value)
1160           (insert (format "(%s . %S)%s"
1161                           key
1162                           strokes
1163                           line-breaking))
1164           (setq attributes (delq key attributes))
1165           )
1166         (setq key (intern (format "%s@%s" 'total-strokes domain)))
1167         (when (and (memq key attributes)
1168                    (setq value (get-char-attribute char key)))
1169           (insert (format "(%s       . %S)%s"
1170                           key
1171                           value
1172                           line-breaking))
1173           (setq attributes (delq key attributes))
1174           )
1175         (dolist (feature '(ideographic-radical
1176                            ideographic-strokes
1177                            total-strokes))
1178           (setq key (intern (format "%s@%s*sources" feature domain)))
1179           (when (and (memq key attributes)
1180                      (setq value (get-char-attribute char key)))
1181             (insert (format "(%s%s" key line-breaking))
1182             (dolist (cell value)
1183               (insert (format " %s" cell)))
1184             (insert ")")
1185             (insert line-breaking)
1186             (setq attributes (delq key attributes))
1187             ))
1188         ))
1189     (when (and (memq 'ideographic-strokes attributes)
1190                (setq value (get-char-attribute char 'ideographic-strokes)))
1191       (setq strokes value)
1192       (insert (format "(ideographic-strokes . %S)%s"
1193                       strokes
1194                       line-breaking))
1195       (setq attributes (delq 'ideographic-strokes attributes))
1196       )
1197     (when (and (memq 'kangxi-radical attributes)
1198                (setq value (get-char-attribute char 'kangxi-radical)))
1199       (unless (eq value radical)
1200         (insert (format "(kangxi-radical\t . %S)\t; %c%s"
1201                         value
1202                         (ideographic-radical value)
1203                         line-breaking))
1204         (or radical
1205             (setq radical value)))
1206       (setq attributes (delq 'kangxi-radical attributes))
1207       )
1208     (when (and (memq 'kangxi-strokes attributes)
1209                (setq value (get-char-attribute char 'kangxi-strokes)))
1210       (unless (eq value strokes)
1211         (insert (format "(kangxi-strokes\t . %S)%s"
1212                         value
1213                         line-breaking))
1214         (or strokes
1215             (setq strokes value)))
1216       (setq attributes (delq 'kangxi-strokes attributes))
1217       )
1218     ;; (when (and (memq 'japanese-radical attributes)
1219     ;;            (setq value (get-char-attribute char 'japanese-radical)))
1220     ;;   (unless (eq value radical)
1221     ;;     (insert (format "(japanese-radical\t . %S)\t; %c%s"
1222     ;;                     value
1223     ;;                     (ideographic-radical value)
1224     ;;                     line-breaking))
1225     ;;     (or radical
1226     ;;         (setq radical value)))
1227     ;;   (setq attributes (delq 'japanese-radical attributes))
1228     ;;   )
1229     (when (and (memq 'japanese-strokes attributes)
1230                (setq value (get-char-attribute char 'japanese-strokes)))
1231       (unless (eq value strokes)
1232         (insert (format "(japanese-strokes\t . %S)%s"
1233                         value
1234                         line-breaking))
1235         (or strokes
1236             (setq strokes value)))
1237       (setq attributes (delq 'japanese-strokes attributes))
1238       )
1239     (when (and (memq 'cns-radical attributes)
1240                (setq value (get-char-attribute char 'cns-radical)))
1241       (insert (format "(cns-radical\t . %S)\t; %c%s"
1242                       value
1243                       (ideographic-radical value)
1244                       line-breaking))
1245       (setq attributes (delq 'cns-radical attributes))
1246       )
1247     (when (and (memq 'cns-strokes attributes)
1248                (setq value (get-char-attribute char 'cns-strokes)))
1249       (unless (eq value strokes)
1250         (insert (format "(cns-strokes\t . %S)%s"
1251                         value
1252                         line-breaking))
1253         (or strokes
1254             (setq strokes value)))
1255       (setq attributes (delq 'cns-strokes attributes))
1256       )
1257     ;; (when (and (memq 'ideographic- attributes)
1258     ;;            (setq value (get-char-attribute char 'ideographic-)))
1259     ;;   (insert "(ideographic-       ")
1260     ;;   (setq lbs (concat "\n" (make-string (current-column) ?\ ))
1261     ;;         separator nil)
1262     ;;   (while (consp value)
1263     ;;     (setq cell (car value))
1264     ;;     (if (integerp cell)
1265     ;;         (setq cell (decode-char '=ucs cell)))
1266     ;;     (cond ((characterp cell)
1267     ;;            (if separator
1268     ;;                (insert lbs))
1269     ;;            (if readable
1270     ;;                (insert (format "%S" cell))
1271     ;;              (char-db-insert-char-spec cell readable))
1272     ;;            (setq separator lbs))
1273     ;;           ((consp cell)
1274     ;;            (if separator
1275     ;;                (insert lbs))
1276     ;;            (if (consp (car cell))
1277     ;;                (char-db-insert-char-spec cell readable)
1278     ;;              (char-db-insert-char-reference cell readable))
1279     ;;            (setq separator lbs))
1280     ;;           (t
1281     ;;            (if separator
1282     ;;                (insert separator))
1283     ;;            (insert (prin1-to-string cell))
1284     ;;            (setq separator " ")))
1285     ;;     (setq value (cdr value)))
1286     ;;   (insert ")")
1287     ;;   (insert line-breaking)
1288     ;;   (setq attributes (delq 'ideographic- attributes)))
1289     (when (and (memq 'total-strokes attributes)
1290                (setq value (get-char-attribute char 'total-strokes)))
1291       (insert (format "(total-strokes       . %S)%s"
1292                       value
1293                       line-breaking))
1294       (setq attributes (delq 'total-strokes attributes))
1295       )
1296     ;; (when (and (memq '->ideograph attributes)
1297     ;;            (setq value (get-char-attribute char '->ideograph)))
1298     ;;   (insert (format "(->ideograph\t%s)%s"
1299     ;;                   (mapconcat (lambda (code)
1300     ;;                                (cond ((symbolp code)
1301     ;;                                       (symbol-name code))
1302     ;;                                      ((integerp code)
1303     ;;                                       (format "#x%04X" code))
1304     ;;                                      (t
1305     ;;                                       (format "%s %S"
1306     ;;                                               line-breaking code))))
1307     ;;                              value " ")
1308     ;;                   line-breaking))
1309     ;;   (setq attributes (delq '->ideograph attributes))
1310     ;;   )
1311     ;; (when (and (memq '->decomposition attributes)
1312     ;;            (setq value (get-char-attribute char '->decomposition)))
1313     ;;   (insert (format "(->decomposition\t%s)%s"
1314     ;;                   (mapconcat (lambda (code)
1315     ;;                                (cond ((symbolp code)
1316     ;;                                       (symbol-name code))
1317     ;;                                      ((characterp code)
1318     ;;                                       (if readable
1319     ;;                                           (format "%S" code)
1320     ;;                                         (format "#x%04X"
1321     ;;                                                 (char-int code))
1322     ;;                                         ))
1323     ;;                                      ((integerp code)
1324     ;;                                       (format "#x%04X" code))
1325     ;;                                      (t
1326     ;;                                       (format "%s%S" line-breaking code))))
1327     ;;                              value " ")
1328     ;;                   line-breaking))
1329     ;;   (setq attributes (delq '->decomposition attributes))
1330     ;;   )
1331     (if (equal (get-char-attribute char '->titlecase)
1332                (get-char-attribute char '->uppercase))
1333         (setq attributes (delq '->titlecase attributes)))
1334     (when (and (memq '->mojikyo attributes)
1335                (setq value (get-char-attribute char '->mojikyo)))
1336       (insert (format "(->mojikyo\t\t. %06d)\t; %c%s"
1337                       value (decode-char 'mojikyo value)
1338                       line-breaking))
1339       (setq attributes (delq '->mojikyo attributes))
1340       )
1341     (unless readable
1342       (dolist (ignored '(composition
1343                          ->denotational <-subsumptive ->ucs-unified
1344                          ;; ->ideographic-component-forms
1345                          ))
1346         (setq attributes (delq ignored attributes))))
1347     (while attributes
1348       (setq name (car attributes))
1349       (unless (eq (setq value (get-char-attribute char name 'value-is-empty))
1350                   'value-is-empty)
1351         (cond ((setq ret (find-charset name))
1352                (setq name (charset-name ret))
1353                (when (not (memq name dest-ccss))
1354                  (setq dest-ccss (cons name dest-ccss))
1355                  (char-db-insert-ccs-feature name value line-breaking))
1356                )
1357               ((string-match "^=>ucs@" (symbol-name name))
1358                (insert (format "(%-18s . #x%04X)\t; %c%s"
1359                                name value (decode-char '=ucs value)
1360                                line-breaking))
1361                )
1362               ((eq name 'jisx0208-1978/4X)
1363                (insert (format "(%-18s . #x%04X)%s"
1364                                name value
1365                                line-breaking))
1366                )
1367               ((and
1368                 (not readable)
1369                 (not (eq name '->subsumptive))
1370                 (not (eq name '->uppercase))
1371                 (not (eq name '->lowercase))
1372                 (not (eq name '->titlecase))
1373                 (not (eq name '->canonical))
1374                 (not (eq name '->Bopomofo))
1375                 (not (eq name '->mistakable))
1376                 (not (eq name '->ideographic-variants))
1377                 (null (get-char-attribute
1378                        char (intern (format "%s*sources" name))))
1379                 (not (string-match "\\*sources$" (symbol-name name)))
1380                 (null (get-char-attribute
1381                        char (intern (format "%s*note" name))))
1382                 (not (string-match "\\*note$" (symbol-name name)))
1383                 (or (eq name '<-identical)
1384                     (eq name '<-uppercase)
1385                     (eq name '<-lowercase)
1386                     (eq name '<-titlecase)
1387                     (eq name '<-canonical)
1388                     (eq name '<-ideographic-variants)
1389                     ;; (eq name '<-synonyms)
1390                     (string-match "^<-synonyms" (symbol-name name))
1391                     (eq name '<-mistakable)
1392                     (when (string-match "^->" (symbol-name name))
1393                       (cond
1394                        ((string-match "^->fullwidth" (symbol-name name))
1395                         (not (and (consp value)
1396                                   (characterp (car value))
1397                                   (encode-char
1398                                    (car value) '=ucs 'defined-only)))
1399                         )
1400                        (t)))
1401                     ))
1402                )
1403               ((or (eq name 'ideographic-structure)
1404                    (eq name 'ideographic-structure@apparent)
1405                    (eq name 'ideographic-combination)
1406                    ;; (eq name 'ideographic-)
1407                    (eq name '=decomposition)
1408                    (char-feature-base-name= '=decomposition name)
1409                    (char-feature-base-name= '=>decomposition name)
1410                    ;; (string-match "^=>*decomposition\\(@[^*]+\\)?$"
1411                    ;;               (symbol-name name))
1412                    (string-match "^\\(->\\|<-\\)[^*]*$" (symbol-name name))
1413                    (string-match "^\\(->\\|<-\\)[^*]*\\*sources$"
1414                                  (symbol-name name))
1415                    )
1416                (char-db-insert-relation-feature char name value
1417                                                 line-breaking
1418                                                 ccss readable))
1419               ;; ((memq name '(ideograph=
1420               ;;               original-ideograph-of
1421               ;;               ancient-ideograph-of
1422               ;;               vulgar-ideograph-of
1423               ;;               wrong-ideograph-of
1424               ;;               ;; simplified-ideograph-of
1425               ;;               ideographic-variants
1426               ;;               ;; ideographic-different-form-of
1427               ;;               ))
1428               ;;  (insert (format "(%-18s%s " name line-breaking))
1429               ;;  (setq lbs (concat "\n" (make-string (current-column) ?\ ))
1430               ;;        separator nil)
1431               ;;  (while (consp value)
1432               ;;    (setq cell (car value))
1433               ;;    (if (and (consp cell)
1434               ;;             (consp (car cell)))
1435               ;;        (progn
1436               ;;          (if separator
1437               ;;              (insert lbs))
1438               ;;          (char-db-insert-alist cell readable)
1439               ;;          (setq separator lbs))
1440               ;;      (if separator
1441               ;;          (insert separator))
1442               ;;      (insert (prin1-to-string cell))
1443               ;;      (setq separator " "))
1444               ;;    (setq value (cdr value)))
1445               ;;  (insert ")")
1446               ;;  (insert line-breaking))
1447               ((consp value)
1448                (insert (format "(%-18s " name))
1449                (setq lbs (concat "\n" (make-string (current-column) ?\ ))
1450                      separator nil)
1451                (while (consp value)
1452                  (setq cell (car value))
1453                  (if (and (consp cell)
1454                           (consp (car cell))
1455                           (setq ret (condition-case nil
1456                                         (find-char cell)
1457                                       (error nil))))
1458                      (progn
1459                        (setq rest cell
1460                              al nil
1461                              cal nil)
1462                        (while rest
1463                          (setq key (car (car rest)))
1464                          (if (find-charset key)
1465                              (setq cal (cons key cal))
1466                            (setq al (cons key al)))
1467                          (setq rest (cdr rest)))
1468                        (if separator
1469                            (insert lbs))
1470                        (insert-char-attributes ret
1471                                                readable
1472                                                al ; cal
1473                                                nil 'for-sub-node)
1474                        (setq separator lbs))
1475                    (setq ret (prin1-to-string cell))
1476                    (if separator
1477                        (if (< (+ (current-column)
1478                                  (length ret)
1479                                  (length separator))
1480                               76)
1481                            (insert separator)
1482                          (insert lbs)))
1483                    (insert ret)
1484                    (setq separator " "))
1485                  (setq value (cdr value)))
1486                (insert ")")
1487                (insert line-breaking))
1488               (t
1489                (insert (format "(%-18s" name))
1490                (setq ret (prin1-to-string value))
1491                (unless (< (+ (current-column)
1492                              (length ret)
1493                              3)
1494                           76)
1495                  (insert line-breaking))
1496                (insert " . " ret ")" line-breaking)
1497                ;; (insert (format "(%-18s . %S)%s"
1498                ;;                 name value
1499                ;;                 line-breaking))
1500                )
1501               ))
1502       (setq attributes (cdr attributes)))
1503     (insert ")")))
1504
1505 (defun insert-char-data (char &optional readable
1506                               attributes)
1507   (save-restriction
1508     (narrow-to-region (point)(point))
1509     (insert "(define-char
1510   '")
1511     (insert-char-attributes char readable attributes)
1512     (insert ")\n")
1513     (goto-char (point-min))
1514     (while (re-search-forward "[ \t]+$" nil t)
1515       (replace-match ""))
1516     ;; from tabify.
1517     (goto-char (point-min))
1518     (while (re-search-forward "[ \t][ \t][ \t]*" nil t)
1519       (let ((column (current-column))
1520             (indent-tabs-mode t))
1521         (delete-region (match-beginning 0) (point))
1522         (indent-to column)))
1523     (goto-char (point-max))
1524     ;; (tabify (point-min)(point-max))
1525     ))
1526
1527 (defun insert-char-data-with-variant (char &optional printable
1528                                            no-ucs-unified
1529                                            script excluded-script)
1530   (insert-char-data char printable)
1531   (let ((variants (char-variants char))
1532         rest
1533         variant vs ret)
1534     (setq variants (sort variants #'<))
1535     (setq rest variants)
1536     (setq variants (cons char variants))
1537     (while rest
1538       (setq variant (car rest))
1539       (unless (get-char-attribute variant '<-subsumptive)
1540         (if (and (or (null script)
1541                      (null (setq vs (get-char-attribute variant 'script)))
1542                      (memq script vs))
1543                  (or (null excluded-script)
1544                      (null (setq vs (get-char-attribute variant 'script)))
1545                      (not (memq excluded-script vs))))
1546             (unless (and no-ucs-unified (get-char-attribute variant '=ucs))
1547               (insert-char-data variant printable)
1548               (if (setq ret (char-variants variant))
1549                   (while ret
1550                     (or (memq (car ret) variants)
1551                         ;; (get-char-attribute (car ret) '<-subsumptive)
1552                         (setq rest (nconc rest (list (car ret)))))
1553                     (setq ret (cdr ret)))))))
1554       (setq rest (cdr rest)))))
1555
1556 (defun insert-char-range-data (min max &optional script excluded-script)
1557   (let ((code min)
1558         char)
1559     (while (<= code max)
1560       (setq char (decode-char '=ucs code))
1561       (if (encode-char char '=ucs 'defined-only)
1562           (insert-char-data-with-variant char nil 'no-ucs-unified
1563                                          script excluded-script))
1564       (setq code (1+ code)))))
1565
1566 (defun write-char-range-data-to-file (min max file
1567                                           &optional script excluded-script)
1568   (let ((coding-system-for-write char-db-file-coding-system))
1569     (with-temp-buffer
1570       (insert (format ";; -*- coding: %s -*-\n"
1571                       char-db-file-coding-system))
1572       (insert-char-range-data min max script excluded-script)
1573       (write-region (point-min)(point-max) file))))
1574
1575 (defvar what-character-original-window-configuration)
1576
1577 ;;;###autoload
1578 (defun what-char-definition (char)
1579   (interactive (list (char-after)))
1580   (let ((buf (get-buffer-create "*Character Description*"))
1581         (the-buf (current-buffer))
1582         (win-conf (current-window-configuration)))
1583     (pop-to-buffer buf)
1584     (make-local-variable 'what-character-original-window-configuration)
1585     (setq what-character-original-window-configuration win-conf)
1586     (setq buffer-read-only nil)
1587     (erase-buffer)
1588     (condition-case err
1589         (progn
1590           (insert-char-data-with-variant char 'printable)
1591           (unless (char-attribute-alist char)
1592             (insert (format ";; = %c\n"
1593                             (let* ((rest (split-char char))
1594                                    (ccs (pop rest))
1595                                    (code (pop rest)))
1596                               (while rest
1597                                 (setq code (logior (lsh code 8)
1598                                                    (pop rest))))
1599                               (decode-char ccs code)))))
1600           ;; (char-db-update-comment)
1601           (set-buffer-modified-p nil)
1602           (view-mode the-buf (lambda (buf)
1603                                (set-window-configuration
1604                                 what-character-original-window-configuration)
1605                                ))
1606           (goto-char (point-min)))
1607       (error (progn
1608                (set-window-configuration
1609                 what-character-original-window-configuration)
1610                (signal (car err) (cdr err)))))))
1611
1612
1613 ;;; @ end
1614 ;;;
1615
1616 (provide 'char-db-util)
1617
1618 ;;; char-db-util.el ends here