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     =daijiten
 199     =daikanwa@rev2
 200     =daikanwa@rev1
 201     =daikanwa/+p
 202     ==daikanwa
 203     ==daijiten
 204     =cbeta
 205     ideograph-hanziku-1
 206     ideograph-hanziku-2
 207     ideograph-hanziku-3
 208     ideograph-hanziku-4
 209     ideograph-hanziku-5
 210     ideograph-hanziku-6
 211     ideograph-hanziku-7
 212     ideograph-hanziku-8
 213     ideograph-hanziku-9
 214     ideograph-hanziku-10
 215     ideograph-hanziku-11
 216     ideograph-hanziku-12
 217     ;; =>>>adobe-japan1-0
 218     ;; =>>>adobe-japan1-1
 219     ;; =>>>adobe-japan1-2
 220     ;; =>>>adobe-japan1-3
 221     ;; =>>>adobe-japan1-4
 222     ;; =>>>adobe-japan1-5
 223     ;; =>>>adobe-japan1-6
 224     ;; =>>>jis-x0208
 225     ;; =>>>jis-x0213-1
 226     ;; =>>>jis-x0213-2
 227     ;; =>>>hanyo-denshi/ja
 228     ;; =>>>hanyo-denshi/jb
 229     ;; =>>>hanyo-denshi/jc
 230     ;; =>>>hanyo-denshi/ft
 231     ;; =>>>hanyo-denshi/ib
 232     ;; =>>>hanyo-denshi/hg
 233     ;; =>>>hanyo-denshi/jt
 234     ;; =>>>hanyo-denshi/ks
 235     ;; =>>>gt
 236     =>>adobe-japan1-0
 237     =>>adobe-japan1-1
 238     =>>adobe-japan1-2
 239     =>>adobe-japan1-3
 240     =>>adobe-japan1-4
 241     =>>adobe-japan1-5
 242     =>>adobe-japan1-6
 243     =>>jis-x0208
 244     =>>jis-x0213-1
 245     =>>jis-x0213-1@2000
 246     =>>jis-x0213-1@2004
 247     =>>jis-x0213-2
 248     =>>jis-x0208@1978
 249     =>>hanyo-denshi/ft
 250     =>>hanyo-denshi/jt
 251     =>>hanyo-denshi/ks
 252     =>>gt
 253     =>>daikanwa
 254     =>>cbeta
 255     =+>jis-x0208
 256     =+>jis-x0213-1
 257     =+>jis-x0213-2
 258     =+>adobe-japan1-0
 259     =+>adobe-japan1-1
 260     =+>adobe-japan1-2
 261     =+>adobe-japan1-3
 262     =+>adobe-japan1-4
 263     =+>adobe-japan1-5
 264     =+>adobe-japan1-6
 265     =+>jis-x0208@1978
 266     =>jis-x0208
 267     =>jis-x0208@1997
 268     =>jis-x0213-1
 269     =>jis-x0213-1@2000
 270     =>jis-x0213-1@2004
 271     =>jis-x0213-2
 272     ==>ucs@bucs
 273     ==>daijiten
 274     =>iwds-1
 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     =>gt
 320     =>mj
 321     =>big5-cdp
 322     =>daikanwa
 323     =>daikanwa/ho
 324     =>cns11643-5
 325     =>cns11643-7
 326     =big5
 327     =big5-eten
 328     =>gt-k
 329     =zinbun-oracle
 330     =>zinbun-oracle
 331     =ruimoku-v6
 332     =>>ruimoku-v6
 333     ==ruimoku-v6
 334     =jef-china3
 335     =>cbeta
 336     =shinjigen
 337     =ucs-var-001
 338     =ucs-var-002
 339     =ucs-var-003
 340     =ucs-var-004
 341     =ucs-var-005
 342     =ucs-var-006
 343     =ucs-var-008
 344     =ucs-var-010
 345     =ucs-itaiji-001
 346     =ucs-itaiji-002
 347     =ucs-itaiji-003
 348     =ucs-itaiji-004
 349     =ucs-itaiji-005
 350     =ucs-itaiji-006
 351     =ucs-itaiji-007
 352     =ucs-itaiji-008
 353     =ucs-itaiji-009
 354     =ucs-itaiji-010
 355     =ucs-itaiji-011
 356     =ucs-itaiji-012
 357     =>ucs-itaiji-001
 358     =>ucs-itaiji-002
 359     =>ucs-itaiji-003
 360     =>ucs-itaiji-004
 361     =>ucs-itaiji-005
 362     =>ucs-itaiji-006
 363     =>ucs-itaiji-007
 364     =>ucs-itaiji-009
 365     =>ucs-itaiji-010
 366     =big5-cdp-var-001
 367     =big5-cdp-var-002
 368     =big5-cdp-var-003
 369     =big5-cdp-var-004
 370     =big5-cdp-var-005
 371     =big5-cdp-var-010
 372     =big5-cdp-itaiji-001
 373     =big5-cdp-itaiji-002
 374     =big5-cdp-itaiji-003
 375     =>big5-cdp-itaiji-001
 376     =>big5-cdp-var-001
 377     =>ucs@iwds-1
 378     =>ucs@cognate
 379     =>ucs@component
 380     =>ucs@iwds-1/normalized
 381     =>ucs-itaiji-001@iwds-1
 382     =>big5-cdp@iwds-1
 383     =>big5-cdp@component
 384     =>big5-cdp@cognate
 385     ==ucs@gb
 386     =ucs@gb
 387     ==ucs-var-002
 388     =ucs@JP/hanazono))
 389
 390
 391 ;;; @ char-db formatters
 392 ;;;
 393
 394 (defun char-db-make-char-spec (char)
 395   (let (ret char-spec)
 396     (cond ((characterp char)
 397            (cond ((and (setq ret (encode-char char '=ucs 'defined-only))
 398                        (not (and (<= #xE000 ret)(<= ret #xF8FF))))
 399                   (setq char-spec (list (cons '=ucs ret)))
 400                   (cond ((setq ret (get-char-attribute char 'name))
 401                          (setq char-spec (cons (cons 'name ret) char-spec))
 402                          )
 403                         ((setq ret (get-char-attribute char 'name*))
 404                          (setq char-spec (cons (cons 'name* ret) char-spec))
 405                          ))
 406                   )
 407                  ((encode-char char '=mj 'defined-only)
 408                   (setq char-spec nil)
 409                   (dolist (ccs (charset-list))
 410                     (if (and (or (eq ccs '=mj)
 411                                  ;; (eq (charset-property ccs 'iso-ir) 177)
 412                                  (string-match "=ucs@" (symbol-name ccs))
 413                                  )
 414                              (setq ccs (charset-name ccs))
 415                              (null (assq ccs char-spec))
 416                              (setq ret (encode-char char ccs 'defined-only)))
 417                         (setq char-spec (cons (cons ccs ret) char-spec))))
 418                   )
 419                  ((encode-char char '==mj 'defined-only)
 420                   (setq char-spec nil)
 421                   (dolist (ccs (charset-list))
 422                     (if (and (or (eq ccs '==mj)
 423                                  ;; (eq (charset-property ccs 'iso-ir) 177)
 424                                  (string-match "=ucs@" (symbol-name ccs))
 425                                  )
 426                              (setq ccs (charset-name ccs))
 427                              (null (assq ccs char-spec))
 428                              (setq ret (encode-char char ccs 'defined-only)))
 429                         (setq char-spec (cons (cons ccs ret) char-spec))))
 430                   )
 431                  ((encode-char char '=adobe-japan1 'defined-only)
 432                   (setq char-spec nil)
 433                   (dolist (ccs (charset-list))
 434                     (if (and (or (memq ccs
 435                                        '(=adobe-japan1-0
 436                                          =adobe-japan1-1
 437                                          =adobe-japan1-2
 438                                          =adobe-japan1-3
 439                                          =adobe-japan1-4
 440                                          =adobe-japan1-5
 441                                          =adobe-japan1-6
 442                                          ))
 443                                  ;; (eq (charset-property ccs 'iso-ir) 177)
 444                                  (string-match "=ucs@" (symbol-name ccs))
 445                                  )
 446                              (setq ccs (charset-name ccs))
 447                              (null (assq ccs char-spec))
 448                              (setq ret (encode-char char ccs 'defined-only)))
 449                         (setq char-spec (cons (cons ccs ret) char-spec))))
 450                   )
 451                  ((encode-char char '==adobe-japan1 'defined-only)
 452                   (setq char-spec nil)
 453                   (dolist (ccs (charset-list))
 454                     (if (and (or (memq ccs
 455                                        '(==adobe-japan1-0
 456                                          ==adobe-japan1-1
 457                                          ==adobe-japan1-2
 458                                          ==adobe-japan1-3
 459                                          ==adobe-japan1-4
 460                                          ==adobe-japan1-5
 461                                          ==adobe-japan1-6
 462                                          ))
 463                                  ;; (eq (charset-property ccs 'iso-ir) 177)
 464                                  (string-match "=ucs@" (symbol-name ccs))
 465                                  )
 466                              (setq ccs (charset-name ccs))
 467                              (null (assq ccs char-spec))
 468                              (setq ret (encode-char char ccs 'defined-only)))
 469                         (setq char-spec (cons (cons ccs ret) char-spec))))
 470                   )
 471                  ((setq ret
 472                         (catch 'tag
 473                           (let ((rest char-db-coded-charset-priority-list)
 474                                 ccs)
 475                             (while rest
 476                               (setq ccs (charset-name
 477                                          (find-charset (car rest))))
 478                               (if (setq ret
 479                                         (encode-char char ccs
 480                                                      'defined-only))
 481                                   (throw 'tag (cons ccs ret)))
 482                               (setq rest (cdr rest))))))
 483                   (setq char-spec (list ret))
 484                   (dolist (ccs (delq (car ret) (charset-list)))
 485                     (if (and (or (charset-iso-final-char ccs)
 486                                  (memq ccs
 487                                        '(=daikanwa
 488                                          =daikanwa@rev2
 489                                          ;; =gt-k
 490                                          =jis-x0208@1997
 491                                          ))
 492                                  (and (eq (charset-property ccs 'iso-ir) 177)
 493                                       (not (eq (charset-name ccs) '=ucs@big5)))
 494                                  ;; (string-match "=ucs@" (symbol-name ccs))
 495                                  )
 496                              (setq ccs (charset-name ccs))
 497                              (null (assq ccs char-spec))
 498                              (setq ret (encode-char char ccs 'defined-only)))
 499                         (setq char-spec (cons (cons ccs ret) char-spec))))
 500                   (if (null char-spec)
 501                       (setq char-spec (split-char char)))
 502                   (cond ((setq ret (get-char-attribute char 'name))
 503                          (setq char-spec (cons (cons 'name ret) char-spec))
 504                          )
 505                         ((setq ret (get-char-attribute char 'name*))
 506                          (setq char-spec (cons (cons 'name* ret) char-spec))
 507                          ))
 508                   )
 509                  ((setq ret (get-char-attribute
 510                              char 'ideographic-combination))
 511                   (setq char-spec
 512                         (cons (cons 'ideographic-combination ret)
 513                               char-spec))
 514                   ))
 515            char-spec)
 516           ((consp char)
 517            char))))
 518
 519 (defun char-db-insert-char-spec (char &optional readable column
 520                                       required-features)
 521   (unless column
 522     (setq column (current-column)))
 523   (let (char-spec temp-char)
 524     (setq char-spec (char-db-make-char-spec char))
 525     (unless (or (characterp char) ; char
 526                 (condition-case nil
 527                     (setq char (find-char char-spec))
 528                   (error nil)))
 529       ;; define temporary character
 530       ;;   Current implementation is dirty.
 531       (setq temp-char (define-char (cons '(ideograph-daikanwa . 0)
 532                                          char-spec)))
 533       (remove-char-attribute temp-char 'ideograph-daikanwa)
 534       (setq char temp-char))
 535     (insert-char-attributes char
 536                             readable
 537                             (union (mapcar #'car char-spec)
 538                                    required-features)
 539                             nil 'for-sub-node)
 540     (when temp-char
 541       ;; undefine temporary character
 542       ;;   Current implementation is dirty.
 543       (setq char-spec (char-attribute-alist temp-char))
 544       (while char-spec
 545         (remove-char-attribute temp-char (car (car char-spec)))
 546         (setq char-spec (cdr char-spec))))))
 547
 548 (defun char-db-insert-alist (alist &optional readable column)
 549   (unless column
 550     (setq column (current-column)))
 551   (let ((line-breaking
 552          (concat "\n" (make-string (1+ column) ?\ )))
 553         name value
 554         ret al ; cal
 555         key
 556         lbs cell rest separator)
 557     (insert "(")
 558     (while alist
 559       (setq name (car (car alist))
 560             value (cdr (car alist)))
 561       (cond ((eq name 'char)
 562              (insert "(char . ")
 563              (if (setq ret (condition-case nil
 564                                (find-char value)
 565                              (error nil)))
 566                  (progn
 567                    (setq al nil
 568                          ;; cal nil
 569                          )
 570                    (while value
 571                      (setq key (car (car value)))
 572                      ;; (if (find-charset key)
 573                      ;;     (setq cal (cons key cal))
 574                      (setq al (cons key al))
 575                      ;; )
 576                      (setq value (cdr value)))
 577                    (insert-char-attributes ret
 578                                            readable
 579                                            (or al 'none) ; cal
 580                                            nil 'for-sub-node))
 581                (insert (prin1-to-string value)))
 582              (insert ")")
 583              (insert line-breaking))
 584             ((consp value)
 585              (insert (format "(%-18s " name))
 586              (setq lbs (concat "\n" (make-string (current-column) ?\ )))
 587              (while (consp value)
 588                (setq cell (car value))
 589                (if (and (consp cell)
 590                         (consp (car cell))
 591                         (setq ret (condition-case nil
 592                                       (find-char cell)
 593                                     (error nil)))
 594                         )
 595                    (progn
 596                      (setq rest cell
 597                            al nil
 598                            ;; cal nil
 599                            )
 600                      (while rest
 601                        (setq key (car (car rest)))
 602                        ;; (if (find-charset key)
 603                        ;;     (setq cal (cons key cal))
 604                        (setq al (cons key al))
 605                        ;; )
 606                        (setq rest (cdr rest)))
 607                      (if separator
 608                          (insert lbs))
 609                      (insert-char-attributes ret
 610                                              readable
 611                                              al ; cal
 612                                              nil 'for-sub-node)
 613                      (setq separator lbs))
 614                  (if separator
 615                      (insert separator))
 616                  (insert (prin1-to-string cell))
 617                  (setq separator " "))
 618                (setq value (cdr value)))
 619              (insert ")")
 620              (insert line-breaking))
 621             (t
 622              (insert (format "(%-18s . %S)%s"
 623                              name value
 624                              line-breaking))))
 625       (setq alist (cdr alist))))
 626   (insert ")"))
 627
 628 (defun char-db-insert-char-reference (plist &optional readable column)
 629   (unless column
 630     (setq column (current-column)))
 631   (let ((line-breaking
 632          (concat "\n" (make-string (1+ column) ?\ )))
 633         (separator "")
 634         name value)
 635     (insert "(")
 636     (while plist
 637       (setq name (pop plist))
 638       (setq value (pop plist))
 639       (cond ((eq name :char)
 640              (insert separator)
 641              (insert ":char\t")
 642              (cond ((numberp value)
 643                     (setq value (decode-char '=ucs value)))
 644                    ;; ((consp value)
 645                    ;;  (setq value (or (find-char value)
 646                    ;;                  value)))
 647                    )
 648              (char-db-insert-char-spec value readable)
 649              (insert line-breaking)
 650              (setq separator ""))
 651             ((eq name :radical)
 652              (insert (format "%s%s\t%d ; %c%s"
 653                              separator
 654                              name value
 655                              (ideographic-radical value)
 656                              line-breaking))
 657              (setq separator ""))
 658             (t
 659              (insert (format "%s%s\t%S" separator name value))
 660              (setq separator line-breaking)))
 661       ))
 662   (insert ")"))
 663
 664 (defun char-db-decode-isolated-char (ccs code-point)
 665   (let (ret)
 666     (setq ret
 667           (cond ((eq ccs 'arabic-iso8859-6)
 668                  (decode-char ccs code-point))
 669                 ;; ((eq ccs '=gt)
 670                 ;;  (decode-builtin-char '==gt code-point))
 671                 ((and (memq ccs '(=gt-pj-1
 672                                   =gt-pj-2
 673                                   =gt-pj-3
 674                                   =gt-pj-4
 675                                   =gt-pj-5
 676                                   =gt-pj-6
 677                                   =gt-pj-7
 678                                   =gt-pj-8
 679                                   =gt-pj-9
 680                                   =gt-pj-10
 681                                   =gt-pj-11))
 682                       (setq ret (decode-char ccs code-point))
 683                       (setq ret (encode-char ret '=gt 'defined-only)))
 684                  (decode-builtin-char '=gt ret))
 685                 (t
 686                  (decode-builtin-char ccs code-point))))
 687     (cond ((null ret)
 688            (or (decode-char ccs code-point)
 689                (define-char (list (cons ccs code-point)))))
 690           ((and (<= 0 (char-int ret))
 691                 (<= (char-int ret) #x1F))
 692            (decode-char '=ucs (+ #x2400 (char-int ret))))
 693           ((= (char-int ret) #x7F)
 694            ?\u2421)
 695           (t ret))))
 696
 697 (defvar char-db-convert-obsolete-format t)
 698
 699 (defun char-db-insert-ccs-feature (name value line-breaking)
 700   (cond
 701    ((integerp value)
 702     (insert
 703      (format
 704       (cond
 705        ((memq name '(=>iwds-1
 706                      ==shinjigen
 707                      =shinjigen
 708                      =shinjigen@1ed ==shinjigen@1ed
 709                      =shinjigen@rev ==shinjigen@rev
 710                      =shinjigen/+p@rev ==shinjigen/+p@rev
 711                      ===daikanwa/ho ==daikanwa/ho
 712                      =daikanwa/ho =>>daikanwa/ho =>daikanwa/ho))
 713         "(%-18s .  %04d)\t; %c")
 714        ((eq name '=shinjigen@1ed/24pr)
 715         "(%-18s . %04d)\t; %c")
 716        ((or
 717          (memq name
 718                '(===daikanwa
 719                  ==daikanwa =daikanwa =>>daikanwa =>daikanwa
 720                  =daikanwa@rev1 =daikanwa@rev2
 721                  =daikanwa/+p ==daikanwa/+p ===daikanwa/+p
 722                  =>>daikanwa/+p
 723                  =daikanwa/+2p =>>daikanwa/+2p
 724                  =gt ==gt ===gt
 725                  =>>gt =+>gt =>gt
 726                  =gt-k ==gt-k ===gt-k
 727                  =>>gt-k =>gt-k
 728                  =adobe-japan1-0 ==adobe-japan1-0 ===adobe-japan1-0
 729                  =adobe-japan1-1 ==adobe-japan1-1 ===adobe-japan1-1
 730                  =adobe-japan1-2 ==adobe-japan1-2 ===adobe-japan1-2
 731                  =adobe-japan1-3 ==adobe-japan1-3 ===adobe-japan1-3
 732                  =adobe-japan1-4 ==adobe-japan1-4 ===adobe-japan1-4
 733                  =adobe-japan1-5 ==adobe-japan1-5 ===adobe-japan1-5
 734                  =adobe-japan1-6 ==adobe-japan1-6 ===adobe-japan1-6
 735                  =>>adobe-japan1-0 =+>adobe-japan1-0
 736                  =>>adobe-japan1-1 =+>adobe-japan1-1
 737                  =>>adobe-japan1-2 =+>adobe-japan1-2
 738                  =>>adobe-japan1-3 =+>adobe-japan1-3
 739                  =>>adobe-japan1-4 =+>adobe-japan1-4
 740                  =>>adobe-japan1-5 =+>adobe-japan1-5
 741                  =>>adobe-japan1-6 =+>adobe-japan1-6
 742                  =>cbeta =cbeta =>>cbeta ==cbeta ===cbeta
 743                  =zinbun-oracle =>zinbun-oracle
 744                  =daijiten ==daijiten ===daijiten ==>daijiten
 745                  ===hng-jou ===hng-keg ===hng-dng ===hng-mam
 746                  ===hng-drt ===hng-kgk ===hng-myz ===hng-kda
 747                  ===hng-khi ===hng-khm ===hng-fhs ===hng-hok
 748                  ===hng-kyd ===hng-sok
 749                  ===hng-yhk ===hng-kak ===hng-kar ===hng-kae
 750                  ===hng-sys ===hng-tsu ===hng-tzj
 751                  ===hng-hos ===hng-kkh ===hng-nak ===hng-jhk
 752                  ===hng-hod ===hng-gok ===hng-ink ===hng-nto
 753                  ===hng-nkm ===hng-k24 ===hng-ini ===hng-nkk
 754                  ===hng-kcc ===hng-kcj ===hng-kbk ===hng-sik
 755                  ===hng-skk ===hng-kyu ===hng-ksk ===hng-wan
 756                  ===hng-okd ===hng-wad ===hng-kmi
 757                  ===hng-zkd ===hng-doh ===hng-jyu ===hng-tzs
 758                  ===hng-sai ===hng-kad ===hng-kss
 759                  ===hng-kyo ===hng-ykk ===hng-saa
 760                  ===hng-sab ===hng-wks ===hng-wke ===hng-smk
 761                  ===hng-sgs ===hng-sts
 762                  =shuowen-jiguge ===shuowen-jiguge4 ===shuowen-jiguge5))
 763          ;; (string-match "^=adobe-" (symbol-name name))
 764          )
 765         "(%-18s . %05d)\t; %c")
 766        ((memq name '(=hanyo-denshi/ks
 767                      ==hanyo-denshi/ks ===hanyo-denshi/ks
 768                      =>>hanyo-denshi/ks
 769                      =koseki ==koseki
 770                      =mj ==mj ===mj =>>mj =>mj
 771                      =zihai mojikyo))
 772         "(%-18s . %06d)\t; %c")
 773        ((memq name '(=hanyo-denshi/tk ==hanyo-denshi/tk))
 774         "(%-18s . %08d)\t; %c")
 775        ((>= (charset-dimension name) 2)
 776         "(%-18s . #x%04X)\t; %c")
 777        (t
 778         "(%-18s . #x%02X)\t; %c"))
 779       name
 780       (if (= (charset-iso-graphic-plane name) 1)
 781           (logior value
 782                   (cond ((= (charset-dimension name) 1)
 783                          #x80)
 784                         ((= (charset-dimension name) 2)
 785                          #x8080)
 786                         ((= (charset-dimension name) 3)
 787                          #x808080)
 788                         (t 0)))
 789         value)
 790       (char-db-decode-isolated-char name value)))
 791     (if (and (= (charset-chars name) 94)
 792              (= (charset-dimension name) 2))
 793         (insert (format " [%02d-%02d]"
 794                         (- (lsh value -8) 32)
 795                         (- (logand value 255) 32))))
 796     )
 797    (t
 798     (insert (format "(%-18s . %s)" name value))
 799     ))
 800   (insert line-breaking))
 801
 802 (defun char-db-insert-relation-feature (char name value line-breaking
 803                                              ccss readable)
 804   (insert (format "(%-18s%s " name line-breaking))
 805   (let ((lbs (concat "\n" (make-string (current-column) ?\ )))
 806         separator cell sources required-features
 807         ret)
 808     (while (consp value)
 809       (setq cell (car value))
 810       (if (integerp cell)
 811           (setq cell (decode-char '=ucs cell)))
 812       (cond
 813        ((eq name '->subsumptive)
 814         (when (or (not (some (lambda (atr)
 815                                (get-char-attribute cell atr))
 816                              char-db-ignored-attributes))
 817                   (some (lambda (ccs)
 818                           (encode-char cell ccs 'defined-only))
 819                         ccss))
 820           (if separator
 821               (insert lbs))
 822           (let ((char-db-ignored-attributes
 823                  (cons '<-subsumptive
 824                        char-db-ignored-attributes)))
 825             (insert-char-attributes cell readable nil nil 'for-sub-node))
 826           (setq separator lbs))
 827         )
 828        ((characterp cell)
 829         (setq sources
 830               (get-char-attribute
 831                char (intern (format "%s*sources" name))))
 832         (setq required-features nil)
 833         (dolist (source sources)
 834           (cond
 835            ((memq source '(JP
 836                            JP/Jouyou
 837                            shinjigen shinjigen@1ed shinjigen@rev))
 838             (setq required-features
 839                   (union required-features
 840                          '(=jis-x0208
 841                            =jis-x0208@1990
 842                            =jis-x0213-1@2000
 843                            =jis-x0213-1@2004
 844                            =jis-x0213-2
 845                            =jis-x0212
 846                            =jis-x0208@1983
 847                            =jis-x0208@1978
 848                            =shinjigen
 849                            =shinjigen@1ed
 850                            =shinjigen@rev
 851                            =shinjigen/+p@rev))))
 852            ((eq source 'CN)
 853             (setq required-features
 854                   (union required-features
 855                          '(=gb2312
 856                            =gb12345
 857                            =iso-ir165)))))
 858           (cond
 859            ((find-charset (setq ret (intern (format "=%s" source))))
 860             (setq required-features
 861                   (cons ret required-features)))
 862            (t (setq required-features
 863                     (cons source required-features)))))
 864         (cond ((string-match "@JP" (symbol-name name))
 865                (setq required-features
 866                      (union required-features
 867                             '(=jis-x0208
 868                               =jis-x0208@1990
 869                               =jis-x0213-1-2000
 870                               =jis-x0213-2-2000
 871                               =jis-x0212
 872                               =jis-x0208@1983
 873                               =jis-x0208@1978))))
 874               ((string-match "@CN" (symbol-name name))
 875                (setq required-features
 876                      (union required-features
 877                             '(=gb2312
 878                               =gb12345
 879                               =iso-ir165)))))
 880         (if separator
 881             (insert lbs))
 882         (if readable
 883             (insert (format "%S" cell))
 884           (char-db-insert-char-spec cell readable
 885                                     nil
 886                                     required-features))
 887         (setq separator lbs))
 888        ((consp cell)
 889         (if separator
 890             (insert lbs))
 891         (if (consp (car cell))
 892             (char-db-insert-char-spec cell readable)
 893           (char-db-insert-char-reference cell readable))
 894         (setq separator lbs))
 895        (t
 896         (if separator
 897             (insert separator))
 898         (insert (prin1-to-string cell))
 899         (setq separator " ")))
 900       (setq value (cdr value)))
 901     (insert ")")
 902     (insert line-breaking)))
 903
 904 (defun insert-char-attributes (char &optional readable attributes column
 905                                     for-sub-node)
 906   (unless column
 907     (setq column (current-column)))
 908   (let (name value ; has-long-ccs-name
 909         rest
 910         radical strokes
 911         (line-breaking
 912          (concat "\n" (make-string (1+ column) ?\ )))
 913         lbs cell separator ret
 914         key al cal
 915         dest-ccss ; sources required-features
 916         ccss)
 917     (let (atr-d)
 918       (setq attributes
 919             (sort (if attributes
 920                       (if (consp attributes)
 921                           (progn
 922                             (dolist (name attributes)
 923                               (unless (memq name char-db-ignored-attributes)
 924                                 (if (find-charset name)
 925                                     (push name ccss))
 926                                 (push name atr-d)))
 927                             atr-d))
 928                     (dolist (name (char-attribute-list))
 929                       (unless (memq name char-db-ignored-attributes)
 930                         (if (find-charset name)
 931                             (push name ccss))
 932                         (push name atr-d)))
 933                     atr-d)
 934                   #'char-attribute-name<)))
 935     (insert "(")
 936     (when (memq '<-subsumptive attributes)
 937       (when (or readable (not for-sub-node))
 938         (when (setq value (get-char-attribute char '<-subsumptive))
 939           (char-db-insert-relation-feature char '<-subsumptive value
 940                                            line-breaking
 941                                            ccss readable)))
 942       (setq attributes (delq '<-subsumptive attributes)))
 943     (when (and (memq '<-denotational attributes)
 944                (setq value (get-char-attribute char '<-denotational)))
 945       (char-db-insert-relation-feature char '<-denotational value
 946                                        line-breaking
 947                                        ccss readable)
 948       (setq attributes (delq '<-denotational attributes)))
 949     (when (and (memq '<-denotational@component attributes)
 950                (setq value (get-char-attribute char '<-denotational@component)))
 951       (char-db-insert-relation-feature char '<-denotational@component value
 952                                        line-breaking
 953                                        ccss readable)
 954       (setq attributes (delq '<-denotational@component attributes)))
 955     (when (and (memq '<-denotational@usage attributes)
 956                (setq value (get-char-attribute char '<-denotational@usage)))
 957       (char-db-insert-relation-feature char '<-denotational@usage value
 958                                        line-breaking
 959                                        ccss readable)
 960       (setq attributes (delq '<-denotational@usage attributes)))
 961     (when (and (memq 'name attributes)
 962                (setq value (get-char-attribute char 'name)))
 963       (insert (format
 964                (if (> (+ (current-column) (length value)) 48)
 965                    "(name . %S)%s"
 966                  "(name               . %S)%s")
 967                value line-breaking))
 968       (setq attributes (delq 'name attributes))
 969       )
 970     (when (and (memq 'name* attributes)
 971                (setq value (get-char-attribute char 'name*)))
 972       (insert (format
 973                (if (> (+ (current-column) (length value)) 48)
 974                    "(name* . %S)%s"
 975                  "(name*              . %S)%s")
 976                value line-breaking))
 977       (setq attributes (delq 'name* attributes))
 978       )
 979     (when (and (memq 'script attributes)
 980                (setq value (get-char-attribute char 'script)))
 981       (insert (format "(script\t\t%s)%s"
 982                       (mapconcat (function prin1-to-string)
 983                                  value " ")
 984                       line-breaking))
 985       (setq attributes (delq 'script attributes))
 986       )
 987     (dolist (name '(=>ucs =>ucs*))
 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 (decode-char '=ucs value)
 992                         line-breaking))
 993         (setq attributes (delq name attributes))))
 994     (dolist (name '(=>ucs@gb =>ucs@big5))
 995       (when (and (memq name attributes)
 996                  (setq value (get-char-attribute char name)))
 997         (insert (format "(%-18s . #x%04X)\t; %c%s"
 998                         name value
 999                         (decode-char (intern
1000                                       (concat "="
1001                                               (substring
1002                                                (symbol-name name) 2)))
1003                                      value)
1004                         line-breaking))
1005         (setq attributes (delq name attributes))
1006         ))
1007     ;; (dolist (name '(=>daikanwa))
1008     ;;   (when (and (memq name attributes)
1009     ;;              (setq value (get-char-attribute char name)))
1010     ;;     (insert
1011     ;;      (if (integerp value)
1012     ;;          (format "(%-18s . %05d)\t; %c%s"
1013     ;;                  name value (decode-char '=daikanwa value)
1014     ;;                  line-breaking)
1015     ;;        (format "(%-18s %s)\t; %c%s"
1016     ;;                name
1017     ;;                (mapconcat (function prin1-to-string)
1018     ;;                           value " ")
1019     ;;                (char-representative-of-daikanwa char)
1020     ;;                line-breaking)))
1021     ;;     (setq attributes (delq name attributes))))
1022     (when (and (memq 'general-category attributes)
1023                (setq value (get-char-attribute char 'general-category)))
1024       (insert (format
1025                "(general-category\t%s) ; %s%s"
1026                (mapconcat (lambda (cell)
1027                             (format "%S" cell))
1028                           value " ")
1029                (cond ((rassoc value unidata-normative-category-alist)
1030                       "Normative Category")
1031                      ((rassoc value unidata-informative-category-alist)
1032                       "Informative Category")
1033                      (t
1034                       "Unknown Category"))
1035                line-breaking))
1036       (setq attributes (delq 'general-category attributes))
1037       )
1038     (when (and (memq 'bidi-category attributes)
1039                (setq value (get-char-attribute char 'bidi-category)))
1040       (insert (format "(bidi-category\t. %S)%s"
1041                       value
1042                       line-breaking))
1043       (setq attributes (delq 'bidi-category attributes))
1044       )
1045     (unless (or (not (memq 'mirrored attributes))
1046                 (eq (setq value (get-char-attribute char 'mirrored 'empty))
1047                     'empty))
1048       (insert (format "(mirrored\t\t. %S)%s"
1049                       value
1050                       line-breaking))
1051       (setq attributes (delq 'mirrored attributes))
1052       )
1053     (cond
1054      ((and (memq 'decimal-digit-value attributes)
1055            (setq value (get-char-attribute char 'decimal-digit-value)))
1056       (insert (format "(decimal-digit-value . %S)%s"
1057                       value
1058                       line-breaking))
1059       (setq attributes (delq 'decimal-digit-value attributes))
1060       (when (and (memq 'digit-value attributes)
1061                  (setq value (get-char-attribute char 'digit-value)))
1062         (insert (format "(digit-value\t . %S)%s"
1063                         value
1064                         line-breaking))
1065         (setq attributes (delq 'digit-value attributes))
1066         )
1067       (when (and (memq 'numeric-value attributes)
1068                  (setq value (get-char-attribute char 'numeric-value)))
1069         (insert (format "(numeric-value\t . %S)%s"
1070                         value
1071                         line-breaking))
1072         (setq attributes (delq 'numeric-value attributes))
1073         )
1074       )
1075      (t
1076       (when (and (memq 'digit-value attributes)
1077                  (setq value (get-char-attribute char 'digit-value)))
1078         (insert (format "(digit-value\t. %S)%s"
1079                         value
1080                         line-breaking))
1081         (setq attributes (delq 'digit-value attributes))
1082         )
1083       (when (and (memq 'numeric-value attributes)
1084                  (setq value (get-char-attribute char 'numeric-value)))
1085         (insert (format "(numeric-value\t. %S)%s"
1086                         value
1087                         line-breaking))
1088         (setq attributes (delq 'numeric-value attributes))
1089         )))
1090     (when (and (memq 'iso-10646-comment attributes)
1091                (setq value (get-char-attribute char 'iso-10646-comment)))
1092       (insert (format "(iso-10646-comment\t. %S)%s"
1093                       value
1094                       line-breaking))
1095       (setq attributes (delq 'iso-10646-comment attributes))
1096       )
1097     (when (and (memq 'morohashi-daikanwa attributes)
1098                (setq value (get-char-attribute char 'morohashi-daikanwa)))
1099       (insert (format "(morohashi-daikanwa\t%s)%s"
1100                       (mapconcat (function prin1-to-string) value " ")
1101                       line-breaking))
1102       (setq attributes (delq 'morohashi-daikanwa attributes))
1103       )
1104     (setq radical nil
1105           strokes nil)
1106     (when (and (memq 'ideographic-radical attributes)
1107                (setq value (get-char-attribute char 'ideographic-radical)))
1108       (setq radical value)
1109       (insert (format "(ideographic-radical . %S)\t; %c%s"
1110                       radical
1111                       (ideographic-radical radical)
1112                       line-breaking))
1113       (setq attributes (delq 'ideographic-radical attributes))
1114       )
1115     (when (and (memq 'shuowen-radical attributes)
1116                (setq value (get-char-attribute char 'shuowen-radical)))
1117       (insert (format "(shuowen-radical\t. %S)\t; %c%s"
1118                       value
1119                       (shuowen-radical value)
1120                       line-breaking))
1121       (setq attributes (delq 'shuowen-radical attributes))
1122       )
1123     (let (key)
1124       (dolist (domain
1125                (append
1126                 char-db-feature-domains
1127                 (let (dest domain)
1128                   (dolist (feature (char-attribute-list))
1129                     (setq feature (symbol-name feature))
1130                     (when (string-match
1131                            "\\(radical\\|strokes\\)@\\([^@*]+\\)\\(\\*\\|$\\)"
1132                            feature)
1133                       (setq domain (intern (match-string 2 feature)))
1134                      (unless (memq domain dest)
1135                        (setq dest (cons domain dest)))))
1136                   (sort dest #'string<))))
1137         (setq key (intern (format "%s@%s" 'ideographic-radical domain)))
1138         (when (and (memq key attributes)
1139                    (setq value (get-char-attribute char key)))
1140           (setq radical value)
1141           (insert (format "(%s . %S)\t; %c%s"
1142                           key
1143                           radical
1144                           (ideographic-radical radical)
1145                           line-breaking))
1146           (setq attributes (delq key attributes))
1147           )
1148         (setq key (intern (format "%s@%s" 'ideographic-strokes domain)))
1149         (when (and (memq key attributes)
1150                    (setq value (get-char-attribute char key)))
1151           (setq strokes value)
1152           (insert (format "(%s . %S)%s"
1153                           key
1154                           strokes
1155                           line-breaking))
1156           (setq attributes (delq key attributes))
1157           )
1158         (setq key (intern (format "%s@%s" 'total-strokes domain)))
1159         (when (and (memq key attributes)
1160                    (setq value (get-char-attribute char key)))
1161           (insert (format "(%s       . %S)%s"
1162                           key
1163                           value
1164                           line-breaking))
1165           (setq attributes (delq key attributes))
1166           )
1167         (dolist (feature '(ideographic-radical
1168                            ideographic-strokes
1169                            total-strokes))
1170           (setq key (intern (format "%s@%s*sources" feature domain)))
1171           (when (and (memq key attributes)
1172                      (setq value (get-char-attribute char key)))
1173             (insert (format "(%s%s" key line-breaking))
1174             (dolist (cell value)
1175               (insert (format " %s" cell)))
1176             (insert ")")
1177             (insert line-breaking)
1178             (setq attributes (delq key attributes))
1179             ))
1180         ))
1181     (when (and (memq 'ideographic-strokes attributes)
1182                (setq value (get-char-attribute char 'ideographic-strokes)))
1183       (setq strokes value)
1184       (insert (format "(ideographic-strokes . %S)%s"
1185                       strokes
1186                       line-breaking))
1187       (setq attributes (delq 'ideographic-strokes attributes))
1188       )
1189     (when (and (memq 'kangxi-radical attributes)
1190                (setq value (get-char-attribute char 'kangxi-radical)))
1191       (unless (eq value radical)
1192         (insert (format "(kangxi-radical\t . %S)\t; %c%s"
1193                         value
1194                         (ideographic-radical value)
1195                         line-breaking))
1196         (or radical
1197             (setq radical value)))
1198       (setq attributes (delq 'kangxi-radical attributes))
1199       )
1200     (when (and (memq 'kangxi-strokes attributes)
1201                (setq value (get-char-attribute char 'kangxi-strokes)))
1202       (unless (eq value strokes)
1203         (insert (format "(kangxi-strokes\t . %S)%s"
1204                         value
1205                         line-breaking))
1206         (or strokes
1207             (setq strokes value)))
1208       (setq attributes (delq 'kangxi-strokes attributes))
1209       )
1210     ;; (when (and (memq 'japanese-radical attributes)
1211     ;;            (setq value (get-char-attribute char 'japanese-radical)))
1212     ;;   (unless (eq value radical)
1213     ;;     (insert (format "(japanese-radical\t . %S)\t; %c%s"
1214     ;;                     value
1215     ;;                     (ideographic-radical value)
1216     ;;                     line-breaking))
1217     ;;     (or radical
1218     ;;         (setq radical value)))
1219     ;;   (setq attributes (delq 'japanese-radical attributes))
1220     ;;   )
1221     (when (and (memq 'japanese-strokes attributes)
1222                (setq value (get-char-attribute char 'japanese-strokes)))
1223       (unless (eq value strokes)
1224         (insert (format "(japanese-strokes\t . %S)%s"
1225                         value
1226                         line-breaking))
1227         (or strokes
1228             (setq strokes value)))
1229       (setq attributes (delq 'japanese-strokes attributes))
1230       )
1231     (when (and (memq 'cns-radical attributes)
1232                (setq value (get-char-attribute char 'cns-radical)))
1233       (insert (format "(cns-radical\t . %S)\t; %c%s"
1234                       value
1235                       (ideographic-radical value)
1236                       line-breaking))
1237       (setq attributes (delq 'cns-radical attributes))
1238       )
1239     (when (and (memq 'cns-strokes attributes)
1240                (setq value (get-char-attribute char 'cns-strokes)))
1241       (unless (eq value strokes)
1242         (insert (format "(cns-strokes\t . %S)%s"
1243                         value
1244                         line-breaking))
1245         (or strokes
1246             (setq strokes value)))
1247       (setq attributes (delq 'cns-strokes attributes))
1248       )
1249     ;; (when (and (memq 'ideographic- attributes)
1250     ;;            (setq value (get-char-attribute char 'ideographic-)))
1251     ;;   (insert "(ideographic-       ")
1252     ;;   (setq lbs (concat "\n" (make-string (current-column) ?\ ))
1253     ;;         separator nil)
1254     ;;   (while (consp value)
1255     ;;     (setq cell (car value))
1256     ;;     (if (integerp cell)
1257     ;;         (setq cell (decode-char '=ucs cell)))
1258     ;;     (cond ((characterp cell)
1259     ;;            (if separator
1260     ;;                (insert lbs))
1261     ;;            (if readable
1262     ;;                (insert (format "%S" cell))
1263     ;;              (char-db-insert-char-spec cell readable))
1264     ;;            (setq separator lbs))
1265     ;;           ((consp cell)
1266     ;;            (if separator
1267     ;;                (insert lbs))
1268     ;;            (if (consp (car cell))
1269     ;;                (char-db-insert-char-spec cell readable)
1270     ;;              (char-db-insert-char-reference cell readable))
1271     ;;            (setq separator lbs))
1272     ;;           (t
1273     ;;            (if separator
1274     ;;                (insert separator))
1275     ;;            (insert (prin1-to-string cell))
1276     ;;            (setq separator " ")))
1277     ;;     (setq value (cdr value)))
1278     ;;   (insert ")")
1279     ;;   (insert line-breaking)
1280     ;;   (setq attributes (delq 'ideographic- attributes)))
1281     (when (and (memq 'total-strokes attributes)
1282                (setq value (get-char-attribute char 'total-strokes)))
1283       (insert (format "(total-strokes       . %S)%s"
1284                       value
1285                       line-breaking))
1286       (setq attributes (delq 'total-strokes attributes))
1287       )
1288     ;; (when (and (memq '->ideograph attributes)
1289     ;;            (setq value (get-char-attribute char '->ideograph)))
1290     ;;   (insert (format "(->ideograph\t%s)%s"
1291     ;;                   (mapconcat (lambda (code)
1292     ;;                                (cond ((symbolp code)
1293     ;;                                       (symbol-name code))
1294     ;;                                      ((integerp code)
1295     ;;                                       (format "#x%04X" code))
1296     ;;                                      (t
1297     ;;                                       (format "%s %S"
1298     ;;                                               line-breaking code))))
1299     ;;                              value " ")
1300     ;;                   line-breaking))
1301     ;;   (setq attributes (delq '->ideograph attributes))
1302     ;;   )
1303     ;; (when (and (memq '->decomposition attributes)
1304     ;;            (setq value (get-char-attribute char '->decomposition)))
1305     ;;   (insert (format "(->decomposition\t%s)%s"
1306     ;;                   (mapconcat (lambda (code)
1307     ;;                                (cond ((symbolp code)
1308     ;;                                       (symbol-name code))
1309     ;;                                      ((characterp code)
1310     ;;                                       (if readable
1311     ;;                                           (format "%S" code)
1312     ;;                                         (format "#x%04X"
1313     ;;                                                 (char-int code))
1314     ;;                                         ))
1315     ;;                                      ((integerp code)
1316     ;;                                       (format "#x%04X" code))
1317     ;;                                      (t
1318     ;;                                       (format "%s%S" line-breaking code))))
1319     ;;                              value " ")
1320     ;;                   line-breaking))
1321     ;;   (setq attributes (delq '->decomposition attributes))
1322     ;;   )
1323     (if (equal (get-char-attribute char '->titlecase)
1324                (get-char-attribute char '->uppercase))
1325         (setq attributes (delq '->titlecase attributes)))
1326     (when (and (memq '->mojikyo attributes)
1327                (setq value (get-char-attribute char '->mojikyo)))
1328       (insert (format "(->mojikyo\t\t. %06d)\t; %c%s"
1329                       value (decode-char 'mojikyo value)
1330                       line-breaking))
1331       (setq attributes (delq '->mojikyo attributes))
1332       )
1333     (unless readable
1334       (dolist (ignored '(composition
1335                          ->denotational <-subsumptive ->ucs-unified
1336                          ;; ->ideographic-component-forms
1337                          ))
1338         (setq attributes (delq ignored attributes))))
1339     (while attributes
1340       (setq name (car attributes))
1341       (unless (eq (setq value (get-char-attribute char name 'value-is-empty))
1342                   'value-is-empty)
1343         (cond ((setq ret (find-charset name))
1344                (setq name (charset-name ret))
1345                (when (not (memq name dest-ccss))
1346                  (setq dest-ccss (cons name dest-ccss))
1347                  (char-db-insert-ccs-feature name value line-breaking))
1348                )
1349               ((string-match "^=>ucs@" (symbol-name name))
1350                (insert (format "(%-18s . #x%04X)\t; %c%s"
1351                                name value (decode-char '=ucs value)
1352                                line-breaking))
1353                )
1354               ((eq name 'jisx0208-1978/4X)
1355                (insert (format "(%-18s . #x%04X)%s"
1356                                name value
1357                                line-breaking))
1358                )
1359               ((and
1360                 (not readable)
1361                 (not (eq name '->subsumptive))
1362                 (not (eq name '->uppercase))
1363                 (not (eq name '->lowercase))
1364                 (not (eq name '->titlecase))
1365                 (not (eq name '->canonical))
1366                 (not (eq name '->Bopomofo))
1367                 (not (eq name '->mistakable))
1368                 (not (eq name '->ideographic-variants))
1369                 (null (get-char-attribute
1370                        char (intern (format "%s*sources" name))))
1371                 (not (string-match "\\*sources$" (symbol-name name)))
1372                 (null (get-char-attribute
1373                        char (intern (format "%s*note" name))))
1374                 (not (string-match "\\*note$" (symbol-name name)))
1375                 (or (eq name '<-identical)
1376                     (eq name '<-uppercase)
1377                     (eq name '<-lowercase)
1378                     (eq name '<-titlecase)
1379                     (eq name '<-canonical)
1380                     (eq name '<-ideographic-variants)
1381                     ;; (eq name '<-synonyms)
1382                     (string-match "^<-synonyms" (symbol-name name))
1383                     (eq name '<-mistakable)
1384                     (when (string-match "^->" (symbol-name name))
1385                       (cond
1386                        ((string-match "^->fullwidth" (symbol-name name))
1387                         (not (and (consp value)
1388                                   (characterp (car value))
1389                                   (encode-char
1390                                    (car value) '=ucs 'defined-only)))
1391                         )
1392                        (t)))
1393                     ))
1394                )
1395               ((or (eq name 'ideographic-structure)
1396                    (eq name 'ideographic-combination)
1397                    ;; (eq name 'ideographic-)
1398                    (eq name '=decomposition)
1399                    (char-feature-base-name= '=decomposition name)
1400                    (char-feature-base-name= '=>decomposition name)
1401                    ;; (string-match "^=>*decomposition\\(@[^*]+\\)?$"
1402                    ;;               (symbol-name name))
1403                    (string-match "^\\(->\\|<-\\)[^*]*$" (symbol-name name))
1404                    (string-match "^\\(->\\|<-\\)[^*]*\\*sources$"
1405                                  (symbol-name name))
1406                    )
1407                (char-db-insert-relation-feature char name value
1408                                                 line-breaking
1409                                                 ccss readable))
1410               ;; ((memq name '(ideograph=
1411               ;;               original-ideograph-of
1412               ;;               ancient-ideograph-of
1413               ;;               vulgar-ideograph-of
1414               ;;               wrong-ideograph-of
1415               ;;               ;; simplified-ideograph-of
1416               ;;               ideographic-variants
1417               ;;               ;; ideographic-different-form-of
1418               ;;               ))
1419               ;;  (insert (format "(%-18s%s " name line-breaking))
1420               ;;  (setq lbs (concat "\n" (make-string (current-column) ?\ ))
1421               ;;        separator nil)
1422               ;;  (while (consp value)
1423               ;;    (setq cell (car value))
1424               ;;    (if (and (consp cell)
1425               ;;             (consp (car cell)))
1426               ;;        (progn
1427               ;;          (if separator
1428               ;;              (insert lbs))
1429               ;;          (char-db-insert-alist cell readable)
1430               ;;          (setq separator lbs))
1431               ;;      (if separator
1432               ;;          (insert separator))
1433               ;;      (insert (prin1-to-string cell))
1434               ;;      (setq separator " "))
1435               ;;    (setq value (cdr value)))
1436               ;;  (insert ")")
1437               ;;  (insert line-breaking))
1438               ((consp value)
1439                (insert (format "(%-18s " name))
1440                (setq lbs (concat "\n" (make-string (current-column) ?\ ))
1441                      separator nil)
1442                (while (consp value)
1443                  (setq cell (car value))
1444                  (if (and (consp cell)
1445                           (consp (car cell))
1446                           (setq ret (condition-case nil
1447                                         (find-char cell)
1448                                       (error nil))))
1449                      (progn
1450                        (setq rest cell
1451                              al nil
1452                              cal nil)
1453                        (while rest
1454                          (setq key (car (car rest)))
1455                          (if (find-charset key)
1456                              (setq cal (cons key cal))
1457                            (setq al (cons key al)))
1458                          (setq rest (cdr rest)))
1459                        (if separator
1460                            (insert lbs))
1461                        (insert-char-attributes ret
1462                                                readable
1463                                                al ; cal
1464                                                nil 'for-sub-node)
1465                        (setq separator lbs))
1466                    (setq ret (prin1-to-string cell))
1467                    (if separator
1468                        (if (< (+ (current-column)
1469                                  (length ret)
1470                                  (length separator))
1471                               76)
1472                            (insert separator)
1473                          (insert lbs)))
1474                    (insert ret)
1475                    (setq separator " "))
1476                  (setq value (cdr value)))
1477                (insert ")")
1478                (insert line-breaking))
1479               (t
1480                (insert (format "(%-18s" name))
1481                (setq ret (prin1-to-string value))
1482                (unless (< (+ (current-column)
1483                              (length ret)
1484                              3)
1485                           76)
1486                  (insert line-breaking))
1487                (insert " . " ret ")" line-breaking)
1488                ;; (insert (format "(%-18s . %S)%s"
1489                ;;                 name value
1490                ;;                 line-breaking))
1491                )
1492               ))
1493       (setq attributes (cdr attributes)))
1494     (insert ")")))
1495
1496 (defun insert-char-data (char &optional readable
1497                               attributes)
1498   (save-restriction
1499     (narrow-to-region (point)(point))
1500     (insert "(define-char
1501   '")
1502     (insert-char-attributes char readable attributes)
1503     (insert ")\n")
1504     (goto-char (point-min))
1505     (while (re-search-forward "[ \t]+$" nil t)
1506       (replace-match ""))
1507     ;; from tabify.
1508     (goto-char (point-min))
1509     (while (re-search-forward "[ \t][ \t][ \t]*" nil t)
1510       (let ((column (current-column))
1511             (indent-tabs-mode t))
1512         (delete-region (match-beginning 0) (point))
1513         (indent-to column)))
1514     (goto-char (point-max))
1515     ;; (tabify (point-min)(point-max))
1516     ))
1517
1518 (defun insert-char-data-with-variant (char &optional printable
1519                                            no-ucs-unified
1520                                            script excluded-script)
1521   (insert-char-data char printable)
1522   (let ((variants (char-variants char))
1523         rest
1524         variant vs ret)
1525     (setq variants (sort variants #'<))
1526     (setq rest variants)
1527     (setq variants (cons char variants))
1528     (while rest
1529       (setq variant (car rest))
1530       (unless (get-char-attribute variant '<-subsumptive)
1531         (if (and (or (null script)
1532                      (null (setq vs (get-char-attribute variant 'script)))
1533                      (memq script vs))
1534                  (or (null excluded-script)
1535                      (null (setq vs (get-char-attribute variant 'script)))
1536                      (not (memq excluded-script vs))))
1537             (unless (and no-ucs-unified (get-char-attribute variant '=ucs))
1538               (insert-char-data variant printable)
1539               (if (setq ret (char-variants variant))
1540                   (while ret
1541                     (or (memq (car ret) variants)
1542                         ;; (get-char-attribute (car ret) '<-subsumptive)
1543                         (setq rest (nconc rest (list (car ret)))))
1544                     (setq ret (cdr ret)))))))
1545       (setq rest (cdr rest)))))
1546
1547 (defun insert-char-range-data (min max &optional script excluded-script)
1548   (let ((code min)
1549         char)
1550     (while (<= code max)
1551       (setq char (decode-char '=ucs code))
1552       (if (encode-char char '=ucs 'defined-only)
1553           (insert-char-data-with-variant char nil 'no-ucs-unified
1554                                          script excluded-script))
1555       (setq code (1+ code)))))
1556
1557 (defun write-char-range-data-to-file (min max file
1558                                           &optional script excluded-script)
1559   (let ((coding-system-for-write char-db-file-coding-system))
1560     (with-temp-buffer
1561       (insert (format ";; -*- coding: %s -*-\n"
1562                       char-db-file-coding-system))
1563       (insert-char-range-data min max script excluded-script)
1564       (write-region (point-min)(point-max) file))))
1565
1566 (defvar what-character-original-window-configuration)
1567
1568 ;;;###autoload
1569 (defun what-char-definition (char)
1570   (interactive (list (char-after)))
1571   (let ((buf (get-buffer-create "*Character Description*"))
1572         (the-buf (current-buffer))
1573         (win-conf (current-window-configuration)))
1574     (pop-to-buffer buf)
1575     (make-local-variable 'what-character-original-window-configuration)
1576     (setq what-character-original-window-configuration win-conf)
1577     (setq buffer-read-only nil)
1578     (erase-buffer)
1579     (condition-case err
1580         (progn
1581           (insert-char-data-with-variant char 'printable)
1582           (unless (char-attribute-alist char)
1583             (insert (format ";; = %c\n"
1584                             (let* ((rest (split-char char))
1585                                    (ccs (pop rest))
1586                                    (code (pop rest)))
1587                               (while rest
1588                                 (setq code (logior (lsh code 8)
1589                                                    (pop rest))))
1590                               (decode-char ccs code)))))
1591           ;; (char-db-update-comment)
1592           (set-buffer-modified-p nil)
1593           (view-mode the-buf (lambda (buf)
1594                                (set-window-configuration
1595                                 what-character-original-window-configuration)
1596                                ))
1597           (goto-char (point-min)))
1598       (error (progn
1599                (set-window-configuration
1600                 what-character-original-window-configuration)
1601                (signal (car err) (cdr err)))))))
1602
1603
1604 ;;; @ end
1605 ;;;
1606
1607 (provide 'char-db-util)
1608
1609 ;;; char-db-util.el ends here