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-itaiji-001
 361     =>ucs-itaiji-002
 362     =>ucs-itaiji-003
 363     =>ucs-itaiji-004
 364     =>ucs-itaiji-005
 365     =>ucs-itaiji-006
 366     =>ucs-itaiji-007
 367     =>ucs-itaiji-008
 368     =>ucs-itaiji-009
 369     =>ucs-itaiji-010
 370     =>ucs-itaiji-011
 371     =big5-cdp-var-001
 372     =big5-cdp-var-002
 373     =big5-cdp-var-003
 374     =big5-cdp-var-004
 375     =big5-cdp-var-005
 376     =big5-cdp-var-007
 377     =big5-cdp-var-010
 378     =big5-cdp-itaiji-001
 379     =big5-cdp-itaiji-002
 380     =big5-cdp-itaiji-003
 381     =>big5-cdp-itaiji-001
 382     =>big5-cdp-itaiji-003
 383     =>big5-cdp-var-001
 384     =>ucs@iwds-1
 385     =>ucs@cognate
 386     =>ucs@component
 387     =>ucs@iwds-1/normalized
 388     =>ucs-itaiji-001@iwds-1
 389     =>ucs-itaiji-002@iwds-1
 390     =>big5-cdp@component
 391     =>big5-cdp@cognate
 392     ==ucs@gb
 393     =ucs@gb
 394     ==ucs-var-002
 395     =ucs@JP/hanazono))
 396
 397
 398 ;;; @ char-db formatters
 399 ;;;
 400
 401 (defun char-db-make-char-spec (char)
 402   (let (ret char-spec)
 403     (cond ((characterp char)
 404            (cond ((and (setq ret (encode-char char '=ucs 'defined-only))
 405                        (not (and (<= #xE000 ret)(<= ret #xF8FF))))
 406                   (setq char-spec (list (cons '=ucs ret)))
 407                   (cond ((setq ret (get-char-attribute char 'name))
 408                          (setq char-spec (cons (cons 'name ret) char-spec))
 409                          )
 410                         ((setq ret (get-char-attribute char 'name*))
 411                          (setq char-spec (cons (cons 'name* ret) char-spec))
 412                          ))
 413                   )
 414                  ((encode-char char '=mj 'defined-only)
 415                   (setq char-spec nil)
 416                   (dolist (ccs (charset-list))
 417                     (if (and (or (eq ccs '=mj)
 418                                  ;; (eq (charset-property ccs 'iso-ir) 177)
 419                                  (string-match "=ucs@" (symbol-name ccs))
 420                                  )
 421                              (setq ccs (charset-name ccs))
 422                              (null (assq ccs char-spec))
 423                              (setq ret (encode-char char ccs 'defined-only)))
 424                         (setq char-spec (cons (cons ccs ret) char-spec))))
 425                   )
 426                  ((encode-char char '==mj 'defined-only)
 427                   (setq char-spec nil)
 428                   (dolist (ccs (charset-list))
 429                     (if (and (or (eq ccs '==mj)
 430                                  ;; (eq (charset-property ccs 'iso-ir) 177)
 431                                  (string-match "=ucs@" (symbol-name ccs))
 432                                  )
 433                              (setq ccs (charset-name ccs))
 434                              (null (assq ccs char-spec))
 435                              (setq ret (encode-char char ccs 'defined-only)))
 436                         (setq char-spec (cons (cons ccs ret) char-spec))))
 437                   )
 438                  ((encode-char char '=adobe-japan1 'defined-only)
 439                   (setq char-spec nil)
 440                   (dolist (ccs (charset-list))
 441                     (if (and (or (memq ccs
 442                                        '(=adobe-japan1-0
 443                                          =adobe-japan1-1
 444                                          =adobe-japan1-2
 445                                          =adobe-japan1-3
 446                                          =adobe-japan1-4
 447                                          =adobe-japan1-5
 448                                          =adobe-japan1-6
 449                                          ))
 450                                  ;; (eq (charset-property ccs 'iso-ir) 177)
 451                                  (string-match "=ucs@" (symbol-name ccs))
 452                                  )
 453                              (setq ccs (charset-name ccs))
 454                              (null (assq ccs char-spec))
 455                              (setq ret (encode-char char ccs 'defined-only)))
 456                         (setq char-spec (cons (cons ccs ret) char-spec))))
 457                   )
 458                  ((encode-char char '==adobe-japan1 'defined-only)
 459                   (setq char-spec nil)
 460                   (dolist (ccs (charset-list))
 461                     (if (and (or (memq ccs
 462                                        '(==adobe-japan1-0
 463                                          ==adobe-japan1-1
 464                                          ==adobe-japan1-2
 465                                          ==adobe-japan1-3
 466                                          ==adobe-japan1-4
 467                                          ==adobe-japan1-5
 468                                          ==adobe-japan1-6
 469                                          ))
 470                                  ;; (eq (charset-property ccs 'iso-ir) 177)
 471                                  (string-match "=ucs@" (symbol-name ccs))
 472                                  )
 473                              (setq ccs (charset-name ccs))
 474                              (null (assq ccs char-spec))
 475                              (setq ret (encode-char char ccs 'defined-only)))
 476                         (setq char-spec (cons (cons ccs ret) char-spec))))
 477                   )
 478                  ((setq ret
 479                         (catch 'tag
 480                           (let ((rest char-db-coded-charset-priority-list)
 481                                 ccs)
 482                             (while rest
 483                               (setq ccs (charset-name
 484                                          (find-charset (car rest))))
 485                               (if (setq ret
 486                                         (encode-char char ccs
 487                                                      'defined-only))
 488                                   (throw 'tag (cons ccs ret)))
 489                               (setq rest (cdr rest))))))
 490                   (setq char-spec (list ret))
 491                   (dolist (ccs (delq (car ret) (charset-list)))
 492                     (if (and (or (charset-iso-final-char ccs)
 493                                  (memq ccs
 494                                        '(=daikanwa
 495                                          =daikanwa@rev2
 496                                          ;; =gt-k
 497                                          =jis-x0208@1997
 498                                          ))
 499                                  (and (eq (charset-property ccs 'iso-ir) 177)
 500                                       (not (eq (charset-name ccs) '=ucs@big5)))
 501                                  ;; (string-match "=ucs@" (symbol-name ccs))
 502                                  )
 503                              (setq ccs (charset-name ccs))
 504                              (null (assq ccs char-spec))
 505                              (setq ret (encode-char char ccs 'defined-only)))
 506                         (setq char-spec (cons (cons ccs ret) char-spec))))
 507                   (if (null char-spec)
 508                       (setq char-spec (split-char char)))
 509                   (cond ((setq ret (get-char-attribute char 'name))
 510                          (setq char-spec (cons (cons 'name ret) char-spec))
 511                          )
 512                         ((setq ret (get-char-attribute char 'name*))
 513                          (setq char-spec (cons (cons 'name* ret) char-spec))
 514                          ))
 515                   )
 516                  ((setq ret (get-char-attribute
 517                              char 'ideographic-combination))
 518                   (setq char-spec
 519                         (cons (cons 'ideographic-combination ret)
 520                               char-spec))
 521                   ))
 522            char-spec)
 523           ((consp char)
 524            char))))
 525
 526 (defun char-db-insert-char-spec (char &optional readable column
 527                                       required-features)
 528   (unless column
 529     (setq column (current-column)))
 530   (let (char-spec temp-char)
 531     (setq char-spec (char-db-make-char-spec char))
 532     (unless (or (characterp char) ; char
 533                 (condition-case nil
 534                     (setq char (find-char char-spec))
 535                   (error nil)))
 536       ;; define temporary character
 537       ;;   Current implementation is dirty.
 538       (setq temp-char (define-char (cons '(ideograph-daikanwa . 0)
 539                                          char-spec)))
 540       (remove-char-attribute temp-char 'ideograph-daikanwa)
 541       (setq char temp-char))
 542     (insert-char-attributes char
 543                             readable
 544                             (union (mapcar #'car char-spec)
 545                                    required-features)
 546                             nil 'for-sub-node)
 547     (when temp-char
 548       ;; undefine temporary character
 549       ;;   Current implementation is dirty.
 550       (setq char-spec (char-attribute-alist temp-char))
 551       (while char-spec
 552         (remove-char-attribute temp-char (car (car char-spec)))
 553         (setq char-spec (cdr char-spec))))))
 554
 555 (defun char-db-insert-alist (alist &optional readable column)
 556   (unless column
 557     (setq column (current-column)))
 558   (let ((line-breaking
 559          (concat "\n" (make-string (1+ column) ?\ )))
 560         name value
 561         ret al ; cal
 562         key
 563         lbs cell rest separator)
 564     (insert "(")
 565     (while alist
 566       (setq name (car (car alist))
 567             value (cdr (car alist)))
 568       (cond ((eq name 'char)
 569              (insert "(char . ")
 570              (if (setq ret (condition-case nil
 571                                (find-char value)
 572                              (error nil)))
 573                  (progn
 574                    (setq al nil
 575                          ;; cal nil
 576                          )
 577                    (while value
 578                      (setq key (car (car value)))
 579                      ;; (if (find-charset key)
 580                      ;;     (setq cal (cons key cal))
 581                      (setq al (cons key al))
 582                      ;; )
 583                      (setq value (cdr value)))
 584                    (insert-char-attributes ret
 585                                            readable
 586                                            (or al 'none) ; cal
 587                                            nil 'for-sub-node))
 588                (insert (prin1-to-string value)))
 589              (insert ")")
 590              (insert line-breaking))
 591             ((consp value)
 592              (insert (format "(%-18s " name))
 593              (setq lbs (concat "\n" (make-string (current-column) ?\ )))
 594              (while (consp value)
 595                (setq cell (car value))
 596                (if (and (consp cell)
 597                         (consp (car cell))
 598                         (setq ret (condition-case nil
 599                                       (find-char cell)
 600                                     (error nil)))
 601                         )
 602                    (progn
 603                      (setq rest cell
 604                            al nil
 605                            ;; cal nil
 606                            )
 607                      (while rest
 608                        (setq key (car (car rest)))
 609                        ;; (if (find-charset key)
 610                        ;;     (setq cal (cons key cal))
 611                        (setq al (cons key al))
 612                        ;; )
 613                        (setq rest (cdr rest)))
 614                      (if separator
 615                          (insert lbs))
 616                      (insert-char-attributes ret
 617                                              readable
 618                                              al ; cal
 619                                              nil 'for-sub-node)
 620                      (setq separator lbs))
 621                  (if separator
 622                      (insert separator))
 623                  (insert (prin1-to-string cell))
 624                  (setq separator " "))
 625                (setq value (cdr value)))
 626              (insert ")")
 627              (insert line-breaking))
 628             (t
 629              (insert (format "(%-18s . %S)%s"
 630                              name value
 631                              line-breaking))))
 632       (setq alist (cdr alist))))
 633   (insert ")"))
 634
 635 (defun char-db-insert-char-reference (plist &optional readable column)
 636   (unless column
 637     (setq column (current-column)))
 638   (let ((line-breaking
 639          (concat "\n" (make-string (1+ column) ?\ )))
 640         (separator "")
 641         name value)
 642     (insert "(")
 643     (while plist
 644       (setq name (pop plist))
 645       (setq value (pop plist))
 646       (cond ((eq name :char)
 647              (insert separator)
 648              (insert ":char\t")
 649              (cond ((numberp value)
 650                     (setq value (decode-char '=ucs value)))
 651                    ;; ((consp value)
 652                    ;;  (setq value (or (find-char value)
 653                    ;;                  value)))
 654                    )
 655              (char-db-insert-char-spec value readable)
 656              (insert line-breaking)
 657              (setq separator ""))
 658             ((eq name :radical)
 659              (insert (format "%s%s\t%d ; %c%s"
 660                              separator
 661                              name value
 662                              (ideographic-radical value)
 663                              line-breaking))
 664              (setq separator ""))
 665             (t
 666              (insert (format "%s%s\t%S" separator name value))
 667              (setq separator line-breaking)))
 668       ))
 669   (insert ")"))
 670
 671 (defun char-db-decode-isolated-char (ccs code-point)
 672   (let (ret)
 673     (setq ret
 674           (cond ((eq ccs 'arabic-iso8859-6)
 675                  (decode-char ccs code-point))
 676                 ;; ((eq ccs '=gt)
 677                 ;;  (decode-builtin-char '==gt code-point))
 678                 ((and (memq ccs '(=gt-pj-1
 679                                   =gt-pj-2
 680                                   =gt-pj-3
 681                                   =gt-pj-4
 682                                   =gt-pj-5
 683                                   =gt-pj-6
 684                                   =gt-pj-7
 685                                   =gt-pj-8
 686                                   =gt-pj-9
 687                                   =gt-pj-10
 688                                   =gt-pj-11))
 689                       (setq ret (decode-char ccs code-point))
 690                       (setq ret (encode-char ret '=gt 'defined-only)))
 691                  (decode-builtin-char '=gt ret))
 692                 (t
 693                  (decode-builtin-char ccs code-point))))
 694     (cond ((null ret)
 695            (or (decode-char ccs code-point)
 696                (define-char (list (cons ccs code-point)))))
 697           ((and (<= 0 (char-int ret))
 698                 (<= (char-int ret) #x1F))
 699            (decode-char '=ucs (+ #x2400 (char-int ret))))
 700           ((= (char-int ret) #x7F)
 701            ?\u2421)
 702           (t ret))))
 703
 704 (defvar char-db-convert-obsolete-format t)
 705
 706 (defun char-db-insert-ccs-feature (name value line-breaking)
 707   (cond
 708    ((integerp value)
 709     (insert
 710      (format
 711       (cond
 712        ((memq name '(=>iwds-1
 713                      ==shinjigen
 714                      =shinjigen
 715                      =shinjigen@1ed ==shinjigen@1ed
 716                      =shinjigen@rev ==shinjigen@rev
 717                      =shinjigen/+p@rev ==shinjigen/+p@rev
 718                      ===daikanwa/ho ==daikanwa/ho
 719                      =daikanwa/ho =>>daikanwa/ho =>daikanwa/ho))
 720         "(%-18s .  %04d)\t; %c")
 721        ((eq name '=shinjigen@1ed/24pr)
 722         "(%-18s . %04d)\t; %c")
 723        ((or
 724          (memq name
 725                '(===daikanwa
 726                  ==daikanwa =daikanwa =>>daikanwa =>daikanwa
 727                  =daikanwa@rev1 =daikanwa@rev2
 728                  =daikanwa/+p ==daikanwa/+p ===daikanwa/+p
 729                  =>>daikanwa/+p
 730                  =daikanwa/+2p =>>daikanwa/+2p
 731                  =gt ==gt ===gt
 732                  =>>gt =+>gt =>gt
 733                  =gt-k ==gt-k ===gt-k
 734                  =>>gt-k =>gt-k
 735                  =adobe-japan1-0 ==adobe-japan1-0 ===adobe-japan1-0
 736                  =adobe-japan1-1 ==adobe-japan1-1 ===adobe-japan1-1
 737                  =adobe-japan1-2 ==adobe-japan1-2 ===adobe-japan1-2
 738                  =adobe-japan1-3 ==adobe-japan1-3 ===adobe-japan1-3
 739                  =adobe-japan1-4 ==adobe-japan1-4 ===adobe-japan1-4
 740                  =adobe-japan1-5 ==adobe-japan1-5 ===adobe-japan1-5
 741                  =adobe-japan1-6 ==adobe-japan1-6 ===adobe-japan1-6
 742                  =>>adobe-japan1-0 =+>adobe-japan1-0
 743                  =>>adobe-japan1-1 =+>adobe-japan1-1
 744                  =>>adobe-japan1-2 =+>adobe-japan1-2
 745                  =>>adobe-japan1-3 =+>adobe-japan1-3
 746                  =>>adobe-japan1-4 =+>adobe-japan1-4
 747                  =>>adobe-japan1-5 =+>adobe-japan1-5
 748                  =>>adobe-japan1-6 =+>adobe-japan1-6
 749                  =>cbeta =cbeta =>>cbeta ==cbeta ===cbeta
 750                  =zinbun-oracle =>zinbun-oracle
 751                  =daijiten ==daijiten ===daijiten ==>daijiten
 752                  ===hng-jou ===hng-keg ===hng-dng ===hng-mam
 753                  ===hng-drt ===hng-kgk ===hng-myz ===hng-kda
 754                  ===hng-khi ===hng-khm ===hng-fhs ===hng-hok
 755                  ===hng-kyd ===hng-sok
 756                  ===hng-yhk ===hng-kak ===hng-kar ===hng-kae
 757                  ===hng-sys ===hng-tsu ===hng-tzj
 758                  ===hng-hos ===hng-kkh ===hng-nak ===hng-jhk
 759                  ===hng-hod ===hng-gok ===hng-ink ===hng-nto
 760                  ===hng-nkm ===hng-k24 ===hng-ini ===hng-nkk
 761                  ===hng-kcc ===hng-kcj ===hng-kbk ===hng-sik
 762                  ===hng-skk ===hng-kyu ===hng-ksk ===hng-wan
 763                  ===hng-okd ===hng-wad ===hng-kmi
 764                  ===hng-zkd ===hng-doh ===hng-jyu ===hng-tzs
 765                  ===hng-sai ===hng-kad ===hng-kss
 766                  ===hng-kyo ===hng-ykk ===hng-saa
 767                  ===hng-sab ===hng-wks ===hng-wke ===hng-smk
 768                  ===hng-sgs ===hng-sts
 769                  =shuowen-jiguge ===shuowen-jiguge4 ===shuowen-jiguge5))
 770          ;; (string-match "^=adobe-" (symbol-name name))
 771          )
 772         "(%-18s . %05d)\t; %c")
 773        ((memq name '(=hanyo-denshi/ks
 774                      ==hanyo-denshi/ks ===hanyo-denshi/ks
 775                      =>>hanyo-denshi/ks
 776                      =koseki ==koseki
 777                      =mj ==mj ===mj =>>mj =>mj
 778                      =zihai mojikyo))
 779         "(%-18s . %06d)\t; %c")
 780        ((memq name '(=hanyo-denshi/tk ==hanyo-denshi/tk))
 781         "(%-18s . %08d)\t; %c")
 782        ((>= (charset-dimension name) 2)
 783         "(%-18s . #x%04X)\t; %c")
 784        (t
 785         "(%-18s . #x%02X)\t; %c"))
 786       name
 787       (if (= (charset-iso-graphic-plane name) 1)
 788           (logior value
 789                   (cond ((= (charset-dimension name) 1)
 790                          #x80)
 791                         ((= (charset-dimension name) 2)
 792                          #x8080)
 793                         ((= (charset-dimension name) 3)
 794                          #x808080)
 795                         (t 0)))
 796         value)
 797       (char-db-decode-isolated-char name value)))
 798     (if (and (= (charset-chars name) 94)
 799              (= (charset-dimension name) 2))
 800         (insert (format " [%02d-%02d]"
 801                         (- (lsh value -8) 32)
 802                         (- (logand value 255) 32))))
 803     )
 804    (t
 805     (insert (format "(%-18s . %s)" name value))
 806     ))
 807   (insert line-breaking))
 808
 809 (defun char-db-insert-relation-feature (char name value line-breaking
 810                                              ccss readable)
 811   (insert (format "(%-18s%s " name line-breaking))
 812   (let ((lbs (concat "\n" (make-string (current-column) ?\ )))
 813         separator cell sources required-features
 814         ret)
 815     (while (consp value)
 816       (setq cell (car value))
 817       (if (integerp cell)
 818           (setq cell (decode-char '=ucs cell)))
 819       (cond
 820        ((eq name '->subsumptive)
 821         (when (or (not (some (lambda (atr)
 822                                (get-char-attribute cell atr))
 823                              char-db-ignored-attributes))
 824                   (some (lambda (ccs)
 825                           (encode-char cell ccs 'defined-only))
 826                         ccss))
 827           (if separator
 828               (insert lbs))
 829           (let ((char-db-ignored-attributes
 830                  (cons '<-subsumptive
 831                        char-db-ignored-attributes)))
 832             (insert-char-attributes cell readable nil nil 'for-sub-node))
 833           (setq separator lbs))
 834         )
 835        ((characterp cell)
 836         (setq sources
 837               (get-char-attribute
 838                char (intern (format "%s*sources" name))))
 839         (setq required-features nil)
 840         (dolist (source sources)
 841           (cond
 842            ((memq source '(JP
 843                            JP/Jouyou
 844                            shinjigen shinjigen@1ed shinjigen@rev))
 845             (setq required-features
 846                   (union required-features
 847                          '(=jis-x0208
 848                            =jis-x0208@1990
 849                            =jis-x0213-1@2000
 850                            =jis-x0213-1@2004
 851                            =jis-x0213-2
 852                            =jis-x0212
 853                            =jis-x0208@1983
 854                            =jis-x0208@1978
 855                            =shinjigen
 856                            =shinjigen@1ed
 857                            =shinjigen@rev
 858                            =shinjigen/+p@rev))))
 859            ((eq source 'CN)
 860             (setq required-features
 861                   (union required-features
 862                          '(=gb2312
 863                            =gb12345
 864                            =iso-ir165)))))
 865           (cond
 866            ((find-charset (setq ret (intern (format "=%s" source))))
 867             (setq required-features
 868                   (cons ret required-features)))
 869            (t (setq required-features
 870                     (cons source required-features)))))
 871         (cond ((string-match "@JP" (symbol-name name))
 872                (setq required-features
 873                      (union required-features
 874                             '(=jis-x0208
 875                               =jis-x0208@1990
 876                               =jis-x0213-1-2000
 877                               =jis-x0213-2-2000
 878                               =jis-x0212
 879                               =jis-x0208@1983
 880                               =jis-x0208@1978))))
 881               ((string-match "@CN" (symbol-name name))
 882                (setq required-features
 883                      (union required-features
 884                             '(=gb2312
 885                               =gb12345
 886                               =iso-ir165)))))
 887         (if separator
 888             (insert lbs))
 889         (if readable
 890             (insert (format "%S" cell))
 891           (char-db-insert-char-spec cell readable
 892                                     nil
 893                                     required-features))
 894         (setq separator lbs))
 895        ((consp cell)
 896         (if separator
 897             (insert lbs))
 898         (if (consp (car cell))
 899             (char-db-insert-char-spec cell readable)
 900           (char-db-insert-char-reference cell readable))
 901         (setq separator lbs))
 902        (t
 903         (if separator
 904             (insert separator))
 905         (insert (prin1-to-string cell))
 906         (setq separator " ")))
 907       (setq value (cdr value)))
 908     (insert ")")
 909     (insert line-breaking)))
 910
 911 (defun insert-char-attributes (char &optional readable attributes column
 912                                     for-sub-node)
 913   (unless column
 914     (setq column (current-column)))
 915   (let (name value ; has-long-ccs-name
 916         rest
 917         radical strokes
 918         (line-breaking
 919          (concat "\n" (make-string (1+ column) ?\ )))
 920         lbs cell separator ret
 921         key al cal
 922         dest-ccss ; sources required-features
 923         ccss)
 924     (let (atr-d)
 925       (setq attributes
 926             (sort (if attributes
 927                       (if (consp attributes)
 928                           (progn
 929                             (dolist (name attributes)
 930                               (unless (memq name char-db-ignored-attributes)
 931                                 (if (find-charset name)
 932                                     (push name ccss))
 933                                 (push name atr-d)))
 934                             atr-d))
 935                     (dolist (name (char-attribute-list))
 936                       (unless (memq name char-db-ignored-attributes)
 937                         (if (find-charset name)
 938                             (push name ccss))
 939                         (push name atr-d)))
 940                     atr-d)
 941                   #'char-attribute-name<)))
 942     (insert "(")
 943     (when (memq '<-subsumptive attributes)
 944       (when (or readable (not for-sub-node))
 945         (when (setq value (get-char-attribute char '<-subsumptive))
 946           (char-db-insert-relation-feature char '<-subsumptive value
 947                                            line-breaking
 948                                            ccss readable)))
 949       (setq attributes (delq '<-subsumptive attributes)))
 950     (when (and (memq '<-denotational attributes)
 951                (setq value (get-char-attribute char '<-denotational)))
 952       (char-db-insert-relation-feature char '<-denotational value
 953                                        line-breaking
 954                                        ccss readable)
 955       (setq attributes (delq '<-denotational attributes)))
 956     (when (and (memq '<-denotational@component attributes)
 957                (setq value (get-char-attribute char '<-denotational@component)))
 958       (char-db-insert-relation-feature char '<-denotational@component value
 959                                        line-breaking
 960                                        ccss readable)
 961       (setq attributes (delq '<-denotational@component attributes)))
 962     (when (and (memq '<-denotational@usage attributes)
 963                (setq value (get-char-attribute char '<-denotational@usage)))
 964       (char-db-insert-relation-feature char '<-denotational@usage value
 965                                        line-breaking
 966                                        ccss readable)
 967       (setq attributes (delq '<-denotational@usage attributes)))
 968     (when (and (memq 'name attributes)
 969                (setq value (get-char-attribute char 'name)))
 970       (insert (format
 971                (if (> (+ (current-column) (length value)) 48)
 972                    "(name . %S)%s"
 973                  "(name               . %S)%s")
 974                value line-breaking))
 975       (setq attributes (delq 'name attributes))
 976       )
 977     (when (and (memq 'name* attributes)
 978                (setq value (get-char-attribute char 'name*)))
 979       (insert (format
 980                (if (> (+ (current-column) (length value)) 48)
 981                    "(name* . %S)%s"
 982                  "(name*              . %S)%s")
 983                value line-breaking))
 984       (setq attributes (delq 'name* attributes))
 985       )
 986     (when (and (memq 'script attributes)
 987                (setq value (get-char-attribute char 'script)))
 988       (insert (format "(script\t\t%s)%s"
 989                       (mapconcat (function prin1-to-string)
 990                                  value " ")
 991                       line-breaking))
 992       (setq attributes (delq 'script attributes))
 993       )
 994     (dolist (name '(=>ucs =>ucs*))
 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 (decode-char '=ucs value)
 999                         line-breaking))
1000         (setq attributes (delq name attributes))))
1001     (dolist (name '(=>ucs@gb =>ucs@big5))
1002       (when (and (memq name attributes)
1003                  (setq value (get-char-attribute char name)))
1004         (insert (format "(%-18s . #x%04X)\t; %c%s"
1005                         name value
1006                         (decode-char (intern
1007                                       (concat "="
1008                                               (substring
1009                                                (symbol-name name) 2)))
1010                                      value)
1011                         line-breaking))
1012         (setq attributes (delq name attributes))
1013         ))
1014     ;; (dolist (name '(=>daikanwa))
1015     ;;   (when (and (memq name attributes)
1016     ;;              (setq value (get-char-attribute char name)))
1017     ;;     (insert
1018     ;;      (if (integerp value)
1019     ;;          (format "(%-18s . %05d)\t; %c%s"
1020     ;;                  name value (decode-char '=daikanwa value)
1021     ;;                  line-breaking)
1022     ;;        (format "(%-18s %s)\t; %c%s"
1023     ;;                name
1024     ;;                (mapconcat (function prin1-to-string)
1025     ;;                           value " ")
1026     ;;                (char-representative-of-daikanwa char)
1027     ;;                line-breaking)))
1028     ;;     (setq attributes (delq name attributes))))
1029     (when (and (memq 'general-category attributes)
1030                (setq value (get-char-attribute char 'general-category)))
1031       (insert (format
1032                "(general-category\t%s) ; %s%s"
1033                (mapconcat (lambda (cell)
1034                             (format "%S" cell))
1035                           value " ")
1036                (cond ((rassoc value unidata-normative-category-alist)
1037                       "Normative Category")
1038                      ((rassoc value unidata-informative-category-alist)
1039                       "Informative Category")
1040                      (t
1041                       "Unknown Category"))
1042                line-breaking))
1043       (setq attributes (delq 'general-category attributes))
1044       )
1045     (when (and (memq 'bidi-category attributes)
1046                (setq value (get-char-attribute char 'bidi-category)))
1047       (insert (format "(bidi-category\t. %S)%s"
1048                       value
1049                       line-breaking))
1050       (setq attributes (delq 'bidi-category attributes))
1051       )
1052     (unless (or (not (memq 'mirrored attributes))
1053                 (eq (setq value (get-char-attribute char 'mirrored 'empty))
1054                     'empty))
1055       (insert (format "(mirrored\t\t. %S)%s"
1056                       value
1057                       line-breaking))
1058       (setq attributes (delq 'mirrored attributes))
1059       )
1060     (cond
1061      ((and (memq 'decimal-digit-value attributes)
1062            (setq value (get-char-attribute char 'decimal-digit-value)))
1063       (insert (format "(decimal-digit-value . %S)%s"
1064                       value
1065                       line-breaking))
1066       (setq attributes (delq 'decimal-digit-value attributes))
1067       (when (and (memq 'digit-value attributes)
1068                  (setq value (get-char-attribute char 'digit-value)))
1069         (insert (format "(digit-value\t . %S)%s"
1070                         value
1071                         line-breaking))
1072         (setq attributes (delq 'digit-value attributes))
1073         )
1074       (when (and (memq 'numeric-value attributes)
1075                  (setq value (get-char-attribute char 'numeric-value)))
1076         (insert (format "(numeric-value\t . %S)%s"
1077                         value
1078                         line-breaking))
1079         (setq attributes (delq 'numeric-value attributes))
1080         )
1081       )
1082      (t
1083       (when (and (memq 'digit-value attributes)
1084                  (setq value (get-char-attribute char 'digit-value)))
1085         (insert (format "(digit-value\t. %S)%s"
1086                         value
1087                         line-breaking))
1088         (setq attributes (delq 'digit-value attributes))
1089         )
1090       (when (and (memq 'numeric-value attributes)
1091                  (setq value (get-char-attribute char 'numeric-value)))
1092         (insert (format "(numeric-value\t. %S)%s"
1093                         value
1094                         line-breaking))
1095         (setq attributes (delq 'numeric-value attributes))
1096         )))
1097     (when (and (memq 'iso-10646-comment attributes)
1098                (setq value (get-char-attribute char 'iso-10646-comment)))
1099       (insert (format "(iso-10646-comment\t. %S)%s"
1100                       value
1101                       line-breaking))
1102       (setq attributes (delq 'iso-10646-comment attributes))
1103       )
1104     (when (and (memq 'morohashi-daikanwa attributes)
1105                (setq value (get-char-attribute char 'morohashi-daikanwa)))
1106       (insert (format "(morohashi-daikanwa\t%s)%s"
1107                       (mapconcat (function prin1-to-string) value " ")
1108                       line-breaking))
1109       (setq attributes (delq 'morohashi-daikanwa attributes))
1110       )
1111     (setq radical nil
1112           strokes nil)
1113     (when (and (memq 'ideographic-radical attributes)
1114                (setq value (get-char-attribute char 'ideographic-radical)))
1115       (setq radical value)
1116       (insert (format "(ideographic-radical . %S)\t; %c%s"
1117                       radical
1118                       (ideographic-radical radical)
1119                       line-breaking))
1120       (setq attributes (delq 'ideographic-radical attributes))
1121       )
1122     (when (and (memq 'shuowen-radical attributes)
1123                (setq value (get-char-attribute char 'shuowen-radical)))
1124       (insert (format "(shuowen-radical\t. %S)\t; %c%s"
1125                       value
1126                       (shuowen-radical value)
1127                       line-breaking))
1128       (setq attributes (delq 'shuowen-radical attributes))
1129       )
1130     (let (key)
1131       (dolist (domain
1132                (append
1133                 char-db-feature-domains
1134                 (let (dest domain)
1135                   (dolist (feature (char-attribute-list))
1136                     (setq feature (symbol-name feature))
1137                     (when (string-match
1138                            "\\(radical\\|strokes\\)@\\([^@*]+\\)\\(\\*\\|$\\)"
1139                            feature)
1140                       (setq domain (intern (match-string 2 feature)))
1141                      (unless (memq domain dest)
1142                        (setq dest (cons domain dest)))))
1143                   (sort dest #'string<))))
1144         (setq key (intern (format "%s@%s" 'ideographic-radical domain)))
1145         (when (and (memq key attributes)
1146                    (setq value (get-char-attribute char key)))
1147           (setq radical value)
1148           (insert (format "(%s . %S)\t; %c%s"
1149                           key
1150                           radical
1151                           (ideographic-radical radical)
1152                           line-breaking))
1153           (setq attributes (delq key attributes))
1154           )
1155         (setq key (intern (format "%s@%s" 'ideographic-strokes domain)))
1156         (when (and (memq key attributes)
1157                    (setq value (get-char-attribute char key)))
1158           (setq strokes value)
1159           (insert (format "(%s . %S)%s"
1160                           key
1161                           strokes
1162                           line-breaking))
1163           (setq attributes (delq key attributes))
1164           )
1165         (setq key (intern (format "%s@%s" 'total-strokes domain)))
1166         (when (and (memq key attributes)
1167                    (setq value (get-char-attribute char key)))
1168           (insert (format "(%s       . %S)%s"
1169                           key
1170                           value
1171                           line-breaking))
1172           (setq attributes (delq key attributes))
1173           )
1174         (dolist (feature '(ideographic-radical
1175                            ideographic-strokes
1176                            total-strokes))
1177           (setq key (intern (format "%s@%s*sources" feature domain)))
1178           (when (and (memq key attributes)
1179                      (setq value (get-char-attribute char key)))
1180             (insert (format "(%s%s" key line-breaking))
1181             (dolist (cell value)
1182               (insert (format " %s" cell)))
1183             (insert ")")
1184             (insert line-breaking)
1185             (setq attributes (delq key attributes))
1186             ))
1187         ))
1188     (when (and (memq 'ideographic-strokes attributes)
1189                (setq value (get-char-attribute char 'ideographic-strokes)))
1190       (setq strokes value)
1191       (insert (format "(ideographic-strokes . %S)%s"
1192                       strokes
1193                       line-breaking))
1194       (setq attributes (delq 'ideographic-strokes attributes))
1195       )
1196     (when (and (memq 'kangxi-radical attributes)
1197                (setq value (get-char-attribute char 'kangxi-radical)))
1198       (unless (eq value radical)
1199         (insert (format "(kangxi-radical\t . %S)\t; %c%s"
1200                         value
1201                         (ideographic-radical value)
1202                         line-breaking))
1203         (or radical
1204             (setq radical value)))
1205       (setq attributes (delq 'kangxi-radical attributes))
1206       )
1207     (when (and (memq 'kangxi-strokes attributes)
1208                (setq value (get-char-attribute char 'kangxi-strokes)))
1209       (unless (eq value strokes)
1210         (insert (format "(kangxi-strokes\t . %S)%s"
1211                         value
1212                         line-breaking))
1213         (or strokes
1214             (setq strokes value)))
1215       (setq attributes (delq 'kangxi-strokes attributes))
1216       )
1217     ;; (when (and (memq 'japanese-radical attributes)
1218     ;;            (setq value (get-char-attribute char 'japanese-radical)))
1219     ;;   (unless (eq value radical)
1220     ;;     (insert (format "(japanese-radical\t . %S)\t; %c%s"
1221     ;;                     value
1222     ;;                     (ideographic-radical value)
1223     ;;                     line-breaking))
1224     ;;     (or radical
1225     ;;         (setq radical value)))
1226     ;;   (setq attributes (delq 'japanese-radical attributes))
1227     ;;   )
1228     (when (and (memq 'japanese-strokes attributes)
1229                (setq value (get-char-attribute char 'japanese-strokes)))
1230       (unless (eq value strokes)
1231         (insert (format "(japanese-strokes\t . %S)%s"
1232                         value
1233                         line-breaking))
1234         (or strokes
1235             (setq strokes value)))
1236       (setq attributes (delq 'japanese-strokes attributes))
1237       )
1238     (when (and (memq 'cns-radical attributes)
1239                (setq value (get-char-attribute char 'cns-radical)))
1240       (insert (format "(cns-radical\t . %S)\t; %c%s"
1241                       value
1242                       (ideographic-radical value)
1243                       line-breaking))
1244       (setq attributes (delq 'cns-radical attributes))
1245       )
1246     (when (and (memq 'cns-strokes attributes)
1247                (setq value (get-char-attribute char 'cns-strokes)))
1248       (unless (eq value strokes)
1249         (insert (format "(cns-strokes\t . %S)%s"
1250                         value
1251                         line-breaking))
1252         (or strokes
1253             (setq strokes value)))
1254       (setq attributes (delq 'cns-strokes attributes))
1255       )
1256     ;; (when (and (memq 'ideographic- attributes)
1257     ;;            (setq value (get-char-attribute char 'ideographic-)))
1258     ;;   (insert "(ideographic-       ")
1259     ;;   (setq lbs (concat "\n" (make-string (current-column) ?\ ))
1260     ;;         separator nil)
1261     ;;   (while (consp value)
1262     ;;     (setq cell (car value))
1263     ;;     (if (integerp cell)
1264     ;;         (setq cell (decode-char '=ucs cell)))
1265     ;;     (cond ((characterp cell)
1266     ;;            (if separator
1267     ;;                (insert lbs))
1268     ;;            (if readable
1269     ;;                (insert (format "%S" cell))
1270     ;;              (char-db-insert-char-spec cell readable))
1271     ;;            (setq separator lbs))
1272     ;;           ((consp cell)
1273     ;;            (if separator
1274     ;;                (insert lbs))
1275     ;;            (if (consp (car cell))
1276     ;;                (char-db-insert-char-spec cell readable)
1277     ;;              (char-db-insert-char-reference cell readable))
1278     ;;            (setq separator lbs))
1279     ;;           (t
1280     ;;            (if separator
1281     ;;                (insert separator))
1282     ;;            (insert (prin1-to-string cell))
1283     ;;            (setq separator " ")))
1284     ;;     (setq value (cdr value)))
1285     ;;   (insert ")")
1286     ;;   (insert line-breaking)
1287     ;;   (setq attributes (delq 'ideographic- attributes)))
1288     (when (and (memq 'total-strokes attributes)
1289                (setq value (get-char-attribute char 'total-strokes)))
1290       (insert (format "(total-strokes       . %S)%s"
1291                       value
1292                       line-breaking))
1293       (setq attributes (delq 'total-strokes attributes))
1294       )
1295     ;; (when (and (memq '->ideograph attributes)
1296     ;;            (setq value (get-char-attribute char '->ideograph)))
1297     ;;   (insert (format "(->ideograph\t%s)%s"
1298     ;;                   (mapconcat (lambda (code)
1299     ;;                                (cond ((symbolp code)
1300     ;;                                       (symbol-name code))
1301     ;;                                      ((integerp code)
1302     ;;                                       (format "#x%04X" code))
1303     ;;                                      (t
1304     ;;                                       (format "%s %S"
1305     ;;                                               line-breaking code))))
1306     ;;                              value " ")
1307     ;;                   line-breaking))
1308     ;;   (setq attributes (delq '->ideograph attributes))
1309     ;;   )
1310     ;; (when (and (memq '->decomposition attributes)
1311     ;;            (setq value (get-char-attribute char '->decomposition)))
1312     ;;   (insert (format "(->decomposition\t%s)%s"
1313     ;;                   (mapconcat (lambda (code)
1314     ;;                                (cond ((symbolp code)
1315     ;;                                       (symbol-name code))
1316     ;;                                      ((characterp code)
1317     ;;                                       (if readable
1318     ;;                                           (format "%S" code)
1319     ;;                                         (format "#x%04X"
1320     ;;                                                 (char-int code))
1321     ;;                                         ))
1322     ;;                                      ((integerp code)
1323     ;;                                       (format "#x%04X" code))
1324     ;;                                      (t
1325     ;;                                       (format "%s%S" line-breaking code))))
1326     ;;                              value " ")
1327     ;;                   line-breaking))
1328     ;;   (setq attributes (delq '->decomposition attributes))
1329     ;;   )
1330     (if (equal (get-char-attribute char '->titlecase)
1331                (get-char-attribute char '->uppercase))
1332         (setq attributes (delq '->titlecase attributes)))
1333     (when (and (memq '->mojikyo attributes)
1334                (setq value (get-char-attribute char '->mojikyo)))
1335       (insert (format "(->mojikyo\t\t. %06d)\t; %c%s"
1336                       value (decode-char 'mojikyo value)
1337                       line-breaking))
1338       (setq attributes (delq '->mojikyo attributes))
1339       )
1340     (unless readable
1341       (dolist (ignored '(composition
1342                          ->denotational <-subsumptive ->ucs-unified
1343                          ;; ->ideographic-component-forms
1344                          ))
1345         (setq attributes (delq ignored attributes))))
1346     (while attributes
1347       (setq name (car attributes))
1348       (unless (eq (setq value (get-char-attribute char name 'value-is-empty))
1349                   'value-is-empty)
1350         (cond ((setq ret (find-charset name))
1351                (setq name (charset-name ret))
1352                (when (not (memq name dest-ccss))
1353                  (setq dest-ccss (cons name dest-ccss))
1354                  (char-db-insert-ccs-feature name value line-breaking))
1355                )
1356               ((string-match "^=>ucs@" (symbol-name name))
1357                (insert (format "(%-18s . #x%04X)\t; %c%s"
1358                                name value (decode-char '=ucs value)
1359                                line-breaking))
1360                )
1361               ((eq name 'jisx0208-1978/4X)
1362                (insert (format "(%-18s . #x%04X)%s"
1363                                name value
1364                                line-breaking))
1365                )
1366               ((and
1367                 (not readable)
1368                 (not (eq name '->subsumptive))
1369                 (not (eq name '->uppercase))
1370                 (not (eq name '->lowercase))
1371                 (not (eq name '->titlecase))
1372                 (not (eq name '->canonical))
1373                 (not (eq name '->Bopomofo))
1374                 (not (eq name '->mistakable))
1375                 (not (eq name '->ideographic-variants))
1376                 (null (get-char-attribute
1377                        char (intern (format "%s*sources" name))))
1378                 (not (string-match "\\*sources$" (symbol-name name)))
1379                 (null (get-char-attribute
1380                        char (intern (format "%s*note" name))))
1381                 (not (string-match "\\*note$" (symbol-name name)))
1382                 (or (eq name '<-identical)
1383                     (eq name '<-uppercase)
1384                     (eq name '<-lowercase)
1385                     (eq name '<-titlecase)
1386                     (eq name '<-canonical)
1387                     (eq name '<-ideographic-variants)
1388                     ;; (eq name '<-synonyms)
1389                     (string-match "^<-synonyms" (symbol-name name))
1390                     (eq name '<-mistakable)
1391                     (when (string-match "^->" (symbol-name name))
1392                       (cond
1393                        ((string-match "^->fullwidth" (symbol-name name))
1394                         (not (and (consp value)
1395                                   (characterp (car value))
1396                                   (encode-char
1397                                    (car value) '=ucs 'defined-only)))
1398                         )
1399                        (t)))
1400                     ))
1401                )
1402               ((or (eq name 'ideographic-structure)
1403                    (eq name 'ideographic-combination)
1404                    ;; (eq name 'ideographic-)
1405                    (eq name '=decomposition)
1406                    (char-feature-base-name= '=decomposition name)
1407                    (char-feature-base-name= '=>decomposition name)
1408                    ;; (string-match "^=>*decomposition\\(@[^*]+\\)?$"
1409                    ;;               (symbol-name name))
1410                    (string-match "^\\(->\\|<-\\)[^*]*$" (symbol-name name))
1411                    (string-match "^\\(->\\|<-\\)[^*]*\\*sources$"
1412                                  (symbol-name name))
1413                    )
1414                (char-db-insert-relation-feature char name value
1415                                                 line-breaking
1416                                                 ccss readable))
1417               ;; ((memq name '(ideograph=
1418               ;;               original-ideograph-of
1419               ;;               ancient-ideograph-of
1420               ;;               vulgar-ideograph-of
1421               ;;               wrong-ideograph-of
1422               ;;               ;; simplified-ideograph-of
1423               ;;               ideographic-variants
1424               ;;               ;; ideographic-different-form-of
1425               ;;               ))
1426               ;;  (insert (format "(%-18s%s " name line-breaking))
1427               ;;  (setq lbs (concat "\n" (make-string (current-column) ?\ ))
1428               ;;        separator nil)
1429               ;;  (while (consp value)
1430               ;;    (setq cell (car value))
1431               ;;    (if (and (consp cell)
1432               ;;             (consp (car cell)))
1433               ;;        (progn
1434               ;;          (if separator
1435               ;;              (insert lbs))
1436               ;;          (char-db-insert-alist cell readable)
1437               ;;          (setq separator lbs))
1438               ;;      (if separator
1439               ;;          (insert separator))
1440               ;;      (insert (prin1-to-string cell))
1441               ;;      (setq separator " "))
1442               ;;    (setq value (cdr value)))
1443               ;;  (insert ")")
1444               ;;  (insert line-breaking))
1445               ((consp value)
1446                (insert (format "(%-18s " name))
1447                (setq lbs (concat "\n" (make-string (current-column) ?\ ))
1448                      separator nil)
1449                (while (consp value)
1450                  (setq cell (car value))
1451                  (if (and (consp cell)
1452                           (consp (car cell))
1453                           (setq ret (condition-case nil
1454                                         (find-char cell)
1455                                       (error nil))))
1456                      (progn
1457                        (setq rest cell
1458                              al nil
1459                              cal nil)
1460                        (while rest
1461                          (setq key (car (car rest)))
1462                          (if (find-charset key)
1463                              (setq cal (cons key cal))
1464                            (setq al (cons key al)))
1465                          (setq rest (cdr rest)))
1466                        (if separator
1467                            (insert lbs))
1468                        (insert-char-attributes ret
1469                                                readable
1470                                                al ; cal
1471                                                nil 'for-sub-node)
1472                        (setq separator lbs))
1473                    (setq ret (prin1-to-string cell))
1474                    (if separator
1475                        (if (< (+ (current-column)
1476                                  (length ret)
1477                                  (length separator))
1478                               76)
1479                            (insert separator)
1480                          (insert lbs)))
1481                    (insert ret)
1482                    (setq separator " "))
1483                  (setq value (cdr value)))
1484                (insert ")")
1485                (insert line-breaking))
1486               (t
1487                (insert (format "(%-18s" name))
1488                (setq ret (prin1-to-string value))
1489                (unless (< (+ (current-column)
1490                              (length ret)
1491                              3)
1492                           76)
1493                  (insert line-breaking))
1494                (insert " . " ret ")" line-breaking)
1495                ;; (insert (format "(%-18s . %S)%s"
1496                ;;                 name value
1497                ;;                 line-breaking))
1498                )
1499               ))
1500       (setq attributes (cdr attributes)))
1501     (insert ")")))
1502
1503 (defun insert-char-data (char &optional readable
1504                               attributes)
1505   (save-restriction
1506     (narrow-to-region (point)(point))
1507     (insert "(define-char
1508   '")
1509     (insert-char-attributes char readable attributes)
1510     (insert ")\n")
1511     (goto-char (point-min))
1512     (while (re-search-forward "[ \t]+$" nil t)
1513       (replace-match ""))
1514     ;; from tabify.
1515     (goto-char (point-min))
1516     (while (re-search-forward "[ \t][ \t][ \t]*" nil t)
1517       (let ((column (current-column))
1518             (indent-tabs-mode t))
1519         (delete-region (match-beginning 0) (point))
1520         (indent-to column)))
1521     (goto-char (point-max))
1522     ;; (tabify (point-min)(point-max))
1523     ))
1524
1525 (defun insert-char-data-with-variant (char &optional printable
1526                                            no-ucs-unified
1527                                            script excluded-script)
1528   (insert-char-data char printable)
1529   (let ((variants (char-variants char))
1530         rest
1531         variant vs ret)
1532     (setq variants (sort variants #'<))
1533     (setq rest variants)
1534     (setq variants (cons char variants))
1535     (while rest
1536       (setq variant (car rest))
1537       (unless (get-char-attribute variant '<-subsumptive)
1538         (if (and (or (null script)
1539                      (null (setq vs (get-char-attribute variant 'script)))
1540                      (memq script vs))
1541                  (or (null excluded-script)
1542                      (null (setq vs (get-char-attribute variant 'script)))
1543                      (not (memq excluded-script vs))))
1544             (unless (and no-ucs-unified (get-char-attribute variant '=ucs))
1545               (insert-char-data variant printable)
1546               (if (setq ret (char-variants variant))
1547                   (while ret
1548                     (or (memq (car ret) variants)
1549                         ;; (get-char-attribute (car ret) '<-subsumptive)
1550                         (setq rest (nconc rest (list (car ret)))))
1551                     (setq ret (cdr ret)))))))
1552       (setq rest (cdr rest)))))
1553
1554 (defun insert-char-range-data (min max &optional script excluded-script)
1555   (let ((code min)
1556         char)
1557     (while (<= code max)
1558       (setq char (decode-char '=ucs code))
1559       (if (encode-char char '=ucs 'defined-only)
1560           (insert-char-data-with-variant char nil 'no-ucs-unified
1561                                          script excluded-script))
1562       (setq code (1+ code)))))
1563
1564 (defun write-char-range-data-to-file (min max file
1565                                           &optional script excluded-script)
1566   (let ((coding-system-for-write char-db-file-coding-system))
1567     (with-temp-buffer
1568       (insert (format ";; -*- coding: %s -*-\n"
1569                       char-db-file-coding-system))
1570       (insert-char-range-data min max script excluded-script)
1571       (write-region (point-min)(point-max) file))))
1572
1573 (defvar what-character-original-window-configuration)
1574
1575 ;;;###autoload
1576 (defun what-char-definition (char)
1577   (interactive (list (char-after)))
1578   (let ((buf (get-buffer-create "*Character Description*"))
1579         (the-buf (current-buffer))
1580         (win-conf (current-window-configuration)))
1581     (pop-to-buffer buf)
1582     (make-local-variable 'what-character-original-window-configuration)
1583     (setq what-character-original-window-configuration win-conf)
1584     (setq buffer-read-only nil)
1585     (erase-buffer)
1586     (condition-case err
1587         (progn
1588           (insert-char-data-with-variant char 'printable)
1589           (unless (char-attribute-alist char)
1590             (insert (format ";; = %c\n"
1591                             (let* ((rest (split-char char))
1592                                    (ccs (pop rest))
1593                                    (code (pop rest)))
1594                               (while rest
1595                                 (setq code (logior (lsh code 8)
1596                                                    (pop rest))))
1597                               (decode-char ccs code)))))
1598           ;; (char-db-update-comment)
1599           (set-buffer-modified-p nil)
1600           (view-mode the-buf (lambda (buf)
1601                                (set-window-configuration
1602                                 what-character-original-window-configuration)
1603                                ))
1604           (goto-char (point-min)))
1605       (error (progn
1606                (set-window-configuration
1607                 what-character-original-window-configuration)
1608                (signal (car err) (cdr err)))))))
1609
1610
1611 ;;; @ end
1612 ;;;
1613
1614 (provide 'char-db-util)
1615
1616 ;;; char-db-util.el ends here