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     =>iwds-1
 276     ;; =>ucs@hanyo-denshi
 277     =>ucs@iso
 278     =>ucs@unicode
 279     =>ucs@jis
 280     =>ucs@cns
 281     =>ucs@ks
 282     =+>ucs@iso
 283     =+>ucs@unicode
 284     =+>ucs@jis
 285     =+>ucs@jis/1990
 286     =+>ucs@cns
 287     =+>ucs@ks
 288     =>>ucs@iso
 289     =>>ucs@unicode
 290     =>>ucs@jis
 291     =>>ucs@cns
 292     =>>>ucs@iso
 293     =>>>ucs@unicode
 294     ==ucs@iso
 295     ==ucs@unicode
 296     ;; ==ucs@cns
 297     ==gb2312
 298     ==ks-x1001
 299     ==cns11643-1
 300     ==cns11643-2
 301     ==cns11643-3
 302     ==cns11643-4
 303     ==cns11643-5
 304     ==cns11643-6
 305     ==cns11643-7
 306     ==gt
 307     ==jis-x0208@1990
 308     ;; ==jis-x0208@1983
 309     ==jis-x0208@1978
 310     ==gt-k
 311     =ucs@iso
 312     =ucs@unicode
 313     =ucs@cns
 314     ==big5-cdp
 315     ==cbeta
 316     =>>big5-cdp
 317     =>>gt-k
 318     =+>gt
 319     =+>big5-cdp
 320     =>gt
 321     =>mj
 322     =>big5-cdp
 323     =>daikanwa
 324     =>daikanwa/ho
 325     =>cns11643-5
 326     =>cns11643-7
 327     =big5
 328     =big5-eten
 329     =>gt-k
 330     =zinbun-oracle
 331     =>zinbun-oracle
 332     =ruimoku-v6
 333     =>>ruimoku-v6
 334     ==ruimoku-v6
 335     =jef-china3
 336     =>cbeta
 337     =shinjigen
 338     =ucs-var-001
 339     =ucs-var-002
 340     =ucs-var-003
 341     =ucs-var-004
 342     =ucs-var-005
 343     =ucs-var-006
 344     =ucs-var-007
 345     =ucs-var-008
 346     =ucs-var-010
 347     =ucs-itaiji-001
 348     =ucs-itaiji-002
 349     =ucs-itaiji-003
 350     =ucs-itaiji-004
 351     =ucs-itaiji-005
 352     =ucs-itaiji-006
 353     =ucs-itaiji-007
 354     =ucs-itaiji-008
 355     =ucs-itaiji-009
 356     =ucs-itaiji-010
 357     =ucs-itaiji-011
 358     =ucs-itaiji-012
 359     =>ucs-itaiji-001
 360     =>ucs-itaiji-002
 361     =>ucs-itaiji-003
 362     =>ucs-itaiji-004
 363     =>ucs-itaiji-005
 364     =>ucs-itaiji-006
 365     =>ucs-itaiji-007
 366     =>ucs-itaiji-008
 367     =>ucs-itaiji-009
 368     =>ucs-itaiji-010
 369     =>ucs-itaiji-011
 370     =big5-cdp-var-001
 371     =big5-cdp-var-002
 372     =big5-cdp-var-003
 373     =big5-cdp-var-004
 374     =big5-cdp-var-005
 375     =big5-cdp-var-010
 376     =big5-cdp-itaiji-001
 377     =big5-cdp-itaiji-002
 378     =big5-cdp-itaiji-003
 379     =>big5-cdp-itaiji-001
 380     =>big5-cdp-itaiji-003
 381     =>big5-cdp-var-001
 382     =>ucs@iwds-1
 383     =>ucs@cognate
 384     =>ucs@component
 385     =>ucs@iwds-1/normalized
 386     =>ucs-itaiji-001@iwds-1
 387     =>ucs-itaiji-002@iwds-1
 388     =>big5-cdp@iwds-1
 389     =>big5-cdp@component
 390     =>big5-cdp@cognate
 391     ==ucs@gb
 392     =ucs@gb
 393     ==ucs-var-002
 394     =ucs@JP/hanazono))
 395
 396
 397 ;;; @ char-db formatters
 398 ;;;
 399
 400 (defun char-db-make-char-spec (char)
 401   (let (ret char-spec)
 402     (cond ((characterp char)
 403            (cond ((and (setq ret (encode-char char '=ucs 'defined-only))
 404                        (not (and (<= #xE000 ret)(<= ret #xF8FF))))
 405                   (setq char-spec (list (cons '=ucs ret)))
 406                   (cond ((setq ret (get-char-attribute char 'name))
 407                          (setq char-spec (cons (cons 'name ret) char-spec))
 408                          )
 409                         ((setq ret (get-char-attribute char 'name*))
 410                          (setq char-spec (cons (cons 'name* ret) char-spec))
 411                          ))
 412                   )
 413                  ((encode-char char '=mj 'defined-only)
 414                   (setq char-spec nil)
 415                   (dolist (ccs (charset-list))
 416                     (if (and (or (eq ccs '=mj)
 417                                  ;; (eq (charset-property ccs 'iso-ir) 177)
 418                                  (string-match "=ucs@" (symbol-name ccs))
 419                                  )
 420                              (setq ccs (charset-name ccs))
 421                              (null (assq ccs char-spec))
 422                              (setq ret (encode-char char ccs 'defined-only)))
 423                         (setq char-spec (cons (cons ccs ret) char-spec))))
 424                   )
 425                  ((encode-char char '==mj 'defined-only)
 426                   (setq char-spec nil)
 427                   (dolist (ccs (charset-list))
 428                     (if (and (or (eq ccs '==mj)
 429                                  ;; (eq (charset-property ccs 'iso-ir) 177)
 430                                  (string-match "=ucs@" (symbol-name ccs))
 431                                  )
 432                              (setq ccs (charset-name ccs))
 433                              (null (assq ccs char-spec))
 434                              (setq ret (encode-char char ccs 'defined-only)))
 435                         (setq char-spec (cons (cons ccs ret) char-spec))))
 436                   )
 437                  ((encode-char char '=adobe-japan1 'defined-only)
 438                   (setq char-spec nil)
 439                   (dolist (ccs (charset-list))
 440                     (if (and (or (memq ccs
 441                                        '(=adobe-japan1-0
 442                                          =adobe-japan1-1
 443                                          =adobe-japan1-2
 444                                          =adobe-japan1-3
 445                                          =adobe-japan1-4
 446                                          =adobe-japan1-5
 447                                          =adobe-japan1-6
 448                                          ))
 449                                  ;; (eq (charset-property ccs 'iso-ir) 177)
 450                                  (string-match "=ucs@" (symbol-name ccs))
 451                                  )
 452                              (setq ccs (charset-name ccs))
 453                              (null (assq ccs char-spec))
 454                              (setq ret (encode-char char ccs 'defined-only)))
 455                         (setq char-spec (cons (cons ccs ret) char-spec))))
 456                   )
 457                  ((encode-char char '==adobe-japan1 'defined-only)
 458                   (setq char-spec nil)
 459                   (dolist (ccs (charset-list))
 460                     (if (and (or (memq ccs
 461                                        '(==adobe-japan1-0
 462                                          ==adobe-japan1-1
 463                                          ==adobe-japan1-2
 464                                          ==adobe-japan1-3
 465                                          ==adobe-japan1-4
 466                                          ==adobe-japan1-5
 467                                          ==adobe-japan1-6
 468                                          ))
 469                                  ;; (eq (charset-property ccs 'iso-ir) 177)
 470                                  (string-match "=ucs@" (symbol-name ccs))
 471                                  )
 472                              (setq ccs (charset-name ccs))
 473                              (null (assq ccs char-spec))
 474                              (setq ret (encode-char char ccs 'defined-only)))
 475                         (setq char-spec (cons (cons ccs ret) char-spec))))
 476                   )
 477                  ((setq ret
 478                         (catch 'tag
 479                           (let ((rest char-db-coded-charset-priority-list)
 480                                 ccs)
 481                             (while rest
 482                               (setq ccs (charset-name
 483                                          (find-charset (car rest))))
 484                               (if (setq ret
 485                                         (encode-char char ccs
 486                                                      'defined-only))
 487                                   (throw 'tag (cons ccs ret)))
 488                               (setq rest (cdr rest))))))
 489                   (setq char-spec (list ret))
 490                   (dolist (ccs (delq (car ret) (charset-list)))
 491                     (if (and (or (charset-iso-final-char ccs)
 492                                  (memq ccs
 493                                        '(=daikanwa
 494                                          =daikanwa@rev2
 495                                          ;; =gt-k
 496                                          =jis-x0208@1997
 497                                          ))
 498                                  (and (eq (charset-property ccs 'iso-ir) 177)
 499                                       (not (eq (charset-name ccs) '=ucs@big5)))
 500                                  ;; (string-match "=ucs@" (symbol-name ccs))
 501                                  )
 502                              (setq ccs (charset-name ccs))
 503                              (null (assq ccs char-spec))
 504                              (setq ret (encode-char char ccs 'defined-only)))
 505                         (setq char-spec (cons (cons ccs ret) char-spec))))
 506                   (if (null char-spec)
 507                       (setq char-spec (split-char char)))
 508                   (cond ((setq ret (get-char-attribute char 'name))
 509                          (setq char-spec (cons (cons 'name ret) char-spec))
 510                          )
 511                         ((setq ret (get-char-attribute char 'name*))
 512                          (setq char-spec (cons (cons 'name* ret) char-spec))
 513                          ))
 514                   )
 515                  ((setq ret (get-char-attribute
 516                              char 'ideographic-combination))
 517                   (setq char-spec
 518                         (cons (cons 'ideographic-combination ret)
 519                               char-spec))
 520                   ))
 521            char-spec)
 522           ((consp char)
 523            char))))
 524
 525 (defun char-db-insert-char-spec (char &optional readable column
 526                                       required-features)
 527   (unless column
 528     (setq column (current-column)))
 529   (let (char-spec temp-char)
 530     (setq char-spec (char-db-make-char-spec char))
 531     (unless (or (characterp char) ; char
 532                 (condition-case nil
 533                     (setq char (find-char char-spec))
 534                   (error nil)))
 535       ;; define temporary character
 536       ;;   Current implementation is dirty.
 537       (setq temp-char (define-char (cons '(ideograph-daikanwa . 0)
 538                                          char-spec)))
 539       (remove-char-attribute temp-char 'ideograph-daikanwa)
 540       (setq char temp-char))
 541     (insert-char-attributes char
 542                             readable
 543                             (union (mapcar #'car char-spec)
 544                                    required-features)
 545                             nil 'for-sub-node)
 546     (when temp-char
 547       ;; undefine temporary character
 548       ;;   Current implementation is dirty.
 549       (setq char-spec (char-attribute-alist temp-char))
 550       (while char-spec
 551         (remove-char-attribute temp-char (car (car char-spec)))
 552         (setq char-spec (cdr char-spec))))))
 553
 554 (defun char-db-insert-alist (alist &optional readable column)
 555   (unless column
 556     (setq column (current-column)))
 557   (let ((line-breaking
 558          (concat "\n" (make-string (1+ column) ?\ )))
 559         name value
 560         ret al ; cal
 561         key
 562         lbs cell rest separator)
 563     (insert "(")
 564     (while alist
 565       (setq name (car (car alist))
 566             value (cdr (car alist)))
 567       (cond ((eq name 'char)
 568              (insert "(char . ")
 569              (if (setq ret (condition-case nil
 570                                (find-char value)
 571                              (error nil)))
 572                  (progn
 573                    (setq al nil
 574                          ;; cal nil
 575                          )
 576                    (while value
 577                      (setq key (car (car value)))
 578                      ;; (if (find-charset key)
 579                      ;;     (setq cal (cons key cal))
 580                      (setq al (cons key al))
 581                      ;; )
 582                      (setq value (cdr value)))
 583                    (insert-char-attributes ret
 584                                            readable
 585                                            (or al 'none) ; cal
 586                                            nil 'for-sub-node))
 587                (insert (prin1-to-string value)))
 588              (insert ")")
 589              (insert line-breaking))
 590             ((consp value)
 591              (insert (format "(%-18s " name))
 592              (setq lbs (concat "\n" (make-string (current-column) ?\ )))
 593              (while (consp value)
 594                (setq cell (car value))
 595                (if (and (consp cell)
 596                         (consp (car cell))
 597                         (setq ret (condition-case nil
 598                                       (find-char cell)
 599                                     (error nil)))
 600                         )
 601                    (progn
 602                      (setq rest cell
 603                            al nil
 604                            ;; cal nil
 605                            )
 606                      (while rest
 607                        (setq key (car (car rest)))
 608                        ;; (if (find-charset key)
 609                        ;;     (setq cal (cons key cal))
 610                        (setq al (cons key al))
 611                        ;; )
 612                        (setq rest (cdr rest)))
 613                      (if separator
 614                          (insert lbs))
 615                      (insert-char-attributes ret
 616                                              readable
 617                                              al ; cal
 618                                              nil 'for-sub-node)
 619                      (setq separator lbs))
 620                  (if separator
 621                      (insert separator))
 622                  (insert (prin1-to-string cell))
 623                  (setq separator " "))
 624                (setq value (cdr value)))
 625              (insert ")")
 626              (insert line-breaking))
 627             (t
 628              (insert (format "(%-18s . %S)%s"
 629                              name value
 630                              line-breaking))))
 631       (setq alist (cdr alist))))
 632   (insert ")"))
 633
 634 (defun char-db-insert-char-reference (plist &optional readable column)
 635   (unless column
 636     (setq column (current-column)))
 637   (let ((line-breaking
 638          (concat "\n" (make-string (1+ column) ?\ )))
 639         (separator "")
 640         name value)
 641     (insert "(")
 642     (while plist
 643       (setq name (pop plist))
 644       (setq value (pop plist))
 645       (cond ((eq name :char)
 646              (insert separator)
 647              (insert ":char\t")
 648              (cond ((numberp value)
 649                     (setq value (decode-char '=ucs value)))
 650                    ;; ((consp value)
 651                    ;;  (setq value (or (find-char value)
 652                    ;;                  value)))
 653                    )
 654              (char-db-insert-char-spec value readable)
 655              (insert line-breaking)
 656              (setq separator ""))
 657             ((eq name :radical)
 658              (insert (format "%s%s\t%d ; %c%s"
 659                              separator
 660                              name value
 661                              (ideographic-radical value)
 662                              line-breaking))
 663              (setq separator ""))
 664             (t
 665              (insert (format "%s%s\t%S" separator name value))
 666              (setq separator line-breaking)))
 667       ))
 668   (insert ")"))
 669
 670 (defun char-db-decode-isolated-char (ccs code-point)
 671   (let (ret)
 672     (setq ret
 673           (cond ((eq ccs 'arabic-iso8859-6)
 674                  (decode-char ccs code-point))
 675                 ;; ((eq ccs '=gt)
 676                 ;;  (decode-builtin-char '==gt code-point))
 677                 ((and (memq ccs '(=gt-pj-1
 678                                   =gt-pj-2
 679                                   =gt-pj-3
 680                                   =gt-pj-4
 681                                   =gt-pj-5
 682                                   =gt-pj-6
 683                                   =gt-pj-7
 684                                   =gt-pj-8
 685                                   =gt-pj-9
 686                                   =gt-pj-10
 687                                   =gt-pj-11))
 688                       (setq ret (decode-char ccs code-point))
 689                       (setq ret (encode-char ret '=gt 'defined-only)))
 690                  (decode-builtin-char '=gt ret))
 691                 (t
 692                  (decode-builtin-char ccs code-point))))
 693     (cond ((null ret)
 694            (or (decode-char ccs code-point)
 695                (define-char (list (cons ccs code-point)))))
 696           ((and (<= 0 (char-int ret))
 697                 (<= (char-int ret) #x1F))
 698            (decode-char '=ucs (+ #x2400 (char-int ret))))
 699           ((= (char-int ret) #x7F)
 700            ?\u2421)
 701           (t ret))))
 702
 703 (defvar char-db-convert-obsolete-format t)
 704
 705 (defun char-db-insert-ccs-feature (name value line-breaking)
 706   (cond
 707    ((integerp value)
 708     (insert
 709      (format
 710       (cond
 711        ((memq name '(=>iwds-1
 712                      ==shinjigen
 713                      =shinjigen
 714                      =shinjigen@1ed ==shinjigen@1ed
 715                      =shinjigen@rev ==shinjigen@rev
 716                      =shinjigen/+p@rev ==shinjigen/+p@rev
 717                      ===daikanwa/ho ==daikanwa/ho
 718                      =daikanwa/ho =>>daikanwa/ho =>daikanwa/ho))
 719         "(%-18s .  %04d)\t; %c")
 720        ((eq name '=shinjigen@1ed/24pr)
 721         "(%-18s . %04d)\t; %c")
 722        ((or
 723          (memq name
 724                '(===daikanwa
 725                  ==daikanwa =daikanwa =>>daikanwa =>daikanwa
 726                  =daikanwa@rev1 =daikanwa@rev2
 727                  =daikanwa/+p ==daikanwa/+p ===daikanwa/+p
 728                  =>>daikanwa/+p
 729                  =daikanwa/+2p =>>daikanwa/+2p
 730                  =gt ==gt ===gt
 731                  =>>gt =+>gt =>gt
 732                  =gt-k ==gt-k ===gt-k
 733                  =>>gt-k =>gt-k
 734                  =adobe-japan1-0 ==adobe-japan1-0 ===adobe-japan1-0
 735                  =adobe-japan1-1 ==adobe-japan1-1 ===adobe-japan1-1
 736                  =adobe-japan1-2 ==adobe-japan1-2 ===adobe-japan1-2
 737                  =adobe-japan1-3 ==adobe-japan1-3 ===adobe-japan1-3
 738                  =adobe-japan1-4 ==adobe-japan1-4 ===adobe-japan1-4
 739                  =adobe-japan1-5 ==adobe-japan1-5 ===adobe-japan1-5
 740                  =adobe-japan1-6 ==adobe-japan1-6 ===adobe-japan1-6
 741                  =>>adobe-japan1-0 =+>adobe-japan1-0
 742                  =>>adobe-japan1-1 =+>adobe-japan1-1
 743                  =>>adobe-japan1-2 =+>adobe-japan1-2
 744                  =>>adobe-japan1-3 =+>adobe-japan1-3
 745                  =>>adobe-japan1-4 =+>adobe-japan1-4
 746                  =>>adobe-japan1-5 =+>adobe-japan1-5
 747                  =>>adobe-japan1-6 =+>adobe-japan1-6
 748                  =>cbeta =cbeta =>>cbeta ==cbeta ===cbeta
 749                  =zinbun-oracle =>zinbun-oracle
 750                  =daijiten ==daijiten ===daijiten ==>daijiten
 751                  ===hng-jou ===hng-keg ===hng-dng ===hng-mam
 752                  ===hng-drt ===hng-kgk ===hng-myz ===hng-kda
 753                  ===hng-khi ===hng-khm ===hng-fhs ===hng-hok
 754                  ===hng-kyd ===hng-sok
 755                  ===hng-yhk ===hng-kak ===hng-kar ===hng-kae
 756                  ===hng-sys ===hng-tsu ===hng-tzj
 757                  ===hng-hos ===hng-kkh ===hng-nak ===hng-jhk
 758                  ===hng-hod ===hng-gok ===hng-ink ===hng-nto
 759                  ===hng-nkm ===hng-k24 ===hng-ini ===hng-nkk
 760                  ===hng-kcc ===hng-kcj ===hng-kbk ===hng-sik
 761                  ===hng-skk ===hng-kyu ===hng-ksk ===hng-wan
 762                  ===hng-okd ===hng-wad ===hng-kmi
 763                  ===hng-zkd ===hng-doh ===hng-jyu ===hng-tzs
 764                  ===hng-sai ===hng-kad ===hng-kss
 765                  ===hng-kyo ===hng-ykk ===hng-saa
 766                  ===hng-sab ===hng-wks ===hng-wke ===hng-smk
 767                  ===hng-sgs ===hng-sts
 768                  =shuowen-jiguge ===shuowen-jiguge4 ===shuowen-jiguge5))
 769          ;; (string-match "^=adobe-" (symbol-name name))
 770          )
 771         "(%-18s . %05d)\t; %c")
 772        ((memq name '(=hanyo-denshi/ks
 773                      ==hanyo-denshi/ks ===hanyo-denshi/ks
 774                      =>>hanyo-denshi/ks
 775                      =koseki ==koseki
 776                      =mj ==mj ===mj =>>mj =>mj
 777                      =zihai mojikyo))
 778         "(%-18s . %06d)\t; %c")
 779        ((memq name '(=hanyo-denshi/tk ==hanyo-denshi/tk))
 780         "(%-18s . %08d)\t; %c")
 781        ((>= (charset-dimension name) 2)
 782         "(%-18s . #x%04X)\t; %c")
 783        (t
 784         "(%-18s . #x%02X)\t; %c"))
 785       name
 786       (if (= (charset-iso-graphic-plane name) 1)
 787           (logior value
 788                   (cond ((= (charset-dimension name) 1)
 789                          #x80)
 790                         ((= (charset-dimension name) 2)
 791                          #x8080)
 792                         ((= (charset-dimension name) 3)
 793                          #x808080)
 794                         (t 0)))
 795         value)
 796       (char-db-decode-isolated-char name value)))
 797     (if (and (= (charset-chars name) 94)
 798              (= (charset-dimension name) 2))
 799         (insert (format " [%02d-%02d]"
 800                         (- (lsh value -8) 32)
 801                         (- (logand value 255) 32))))
 802     )
 803    (t
 804     (insert (format "(%-18s . %s)" name value))
 805     ))
 806   (insert line-breaking))
 807
 808 (defun char-db-insert-relation-feature (char name value line-breaking
 809                                              ccss readable)
 810   (insert (format "(%-18s%s " name line-breaking))
 811   (let ((lbs (concat "\n" (make-string (current-column) ?\ )))
 812         separator cell sources required-features
 813         ret)
 814     (while (consp value)
 815       (setq cell (car value))
 816       (if (integerp cell)
 817           (setq cell (decode-char '=ucs cell)))
 818       (cond
 819        ((eq name '->subsumptive)
 820         (when (or (not (some (lambda (atr)
 821                                (get-char-attribute cell atr))
 822                              char-db-ignored-attributes))
 823                   (some (lambda (ccs)
 824                           (encode-char cell ccs 'defined-only))
 825                         ccss))
 826           (if separator
 827               (insert lbs))
 828           (let ((char-db-ignored-attributes
 829                  (cons '<-subsumptive
 830                        char-db-ignored-attributes)))
 831             (insert-char-attributes cell readable nil nil 'for-sub-node))
 832           (setq separator lbs))
 833         )
 834        ((characterp cell)
 835         (setq sources
 836               (get-char-attribute
 837                char (intern (format "%s*sources" name))))
 838         (setq required-features nil)
 839         (dolist (source sources)
 840           (cond
 841            ((memq source '(JP
 842                            JP/Jouyou
 843                            shinjigen shinjigen@1ed shinjigen@rev))
 844             (setq required-features
 845                   (union required-features
 846                          '(=jis-x0208
 847                            =jis-x0208@1990
 848                            =jis-x0213-1@2000
 849                            =jis-x0213-1@2004
 850                            =jis-x0213-2
 851                            =jis-x0212
 852                            =jis-x0208@1983
 853                            =jis-x0208@1978
 854                            =shinjigen
 855                            =shinjigen@1ed
 856                            =shinjigen@rev
 857                            =shinjigen/+p@rev))))
 858            ((eq source 'CN)
 859             (setq required-features
 860                   (union required-features
 861                          '(=gb2312
 862                            =gb12345
 863                            =iso-ir165)))))
 864           (cond
 865            ((find-charset (setq ret (intern (format "=%s" source))))
 866             (setq required-features
 867                   (cons ret required-features)))
 868            (t (setq required-features
 869                     (cons source required-features)))))
 870         (cond ((string-match "@JP" (symbol-name name))
 871                (setq required-features
 872                      (union required-features
 873                             '(=jis-x0208
 874                               =jis-x0208@1990
 875                               =jis-x0213-1-2000
 876                               =jis-x0213-2-2000
 877                               =jis-x0212
 878                               =jis-x0208@1983
 879                               =jis-x0208@1978))))
 880               ((string-match "@CN" (symbol-name name))
 881                (setq required-features
 882                      (union required-features
 883                             '(=gb2312
 884                               =gb12345
 885                               =iso-ir165)))))
 886         (if separator
 887             (insert lbs))
 888         (if readable
 889             (insert (format "%S" cell))
 890           (char-db-insert-char-spec cell readable
 891                                     nil
 892                                     required-features))
 893         (setq separator lbs))
 894        ((consp cell)
 895         (if separator
 896             (insert lbs))
 897         (if (consp (car cell))
 898             (char-db-insert-char-spec cell readable)
 899           (char-db-insert-char-reference cell readable))
 900         (setq separator lbs))
 901        (t
 902         (if separator
 903             (insert separator))
 904         (insert (prin1-to-string cell))
 905         (setq separator " ")))
 906       (setq value (cdr value)))
 907     (insert ")")
 908     (insert line-breaking)))
 909
 910 (defun insert-char-attributes (char &optional readable attributes column
 911                                     for-sub-node)
 912   (unless column
 913     (setq column (current-column)))
 914   (let (name value ; has-long-ccs-name
 915         rest
 916         radical strokes
 917         (line-breaking
 918          (concat "\n" (make-string (1+ column) ?\ )))
 919         lbs cell separator ret
 920         key al cal
 921         dest-ccss ; sources required-features
 922         ccss)
 923     (let (atr-d)
 924       (setq attributes
 925             (sort (if attributes
 926                       (if (consp attributes)
 927                           (progn
 928                             (dolist (name attributes)
 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                     (dolist (name (char-attribute-list))
 935                       (unless (memq name char-db-ignored-attributes)
 936                         (if (find-charset name)
 937                             (push name ccss))
 938                         (push name atr-d)))
 939                     atr-d)
 940                   #'char-attribute-name<)))
 941     (insert "(")
 942     (when (memq '<-subsumptive attributes)
 943       (when (or readable (not for-sub-node))
 944         (when (setq value (get-char-attribute char '<-subsumptive))
 945           (char-db-insert-relation-feature char '<-subsumptive value
 946                                            line-breaking
 947                                            ccss readable)))
 948       (setq attributes (delq '<-subsumptive attributes)))
 949     (when (and (memq '<-denotational attributes)
 950                (setq value (get-char-attribute char '<-denotational)))
 951       (char-db-insert-relation-feature char '<-denotational value
 952                                        line-breaking
 953                                        ccss readable)
 954       (setq attributes (delq '<-denotational attributes)))
 955     (when (and (memq '<-denotational@component attributes)
 956                (setq value (get-char-attribute char '<-denotational@component)))
 957       (char-db-insert-relation-feature char '<-denotational@component value
 958                                        line-breaking
 959                                        ccss readable)
 960       (setq attributes (delq '<-denotational@component attributes)))
 961     (when (and (memq '<-denotational@usage attributes)
 962                (setq value (get-char-attribute char '<-denotational@usage)))
 963       (char-db-insert-relation-feature char '<-denotational@usage value
 964                                        line-breaking
 965                                        ccss readable)
 966       (setq attributes (delq '<-denotational@usage attributes)))
 967     (when (and (memq 'name attributes)
 968                (setq value (get-char-attribute char 'name)))
 969       (insert (format
 970                (if (> (+ (current-column) (length value)) 48)
 971                    "(name . %S)%s"
 972                  "(name               . %S)%s")
 973                value line-breaking))
 974       (setq attributes (delq 'name attributes))
 975       )
 976     (when (and (memq 'name* attributes)
 977                (setq value (get-char-attribute char 'name*)))
 978       (insert (format
 979                (if (> (+ (current-column) (length value)) 48)
 980                    "(name* . %S)%s"
 981                  "(name*              . %S)%s")
 982                value line-breaking))
 983       (setq attributes (delq 'name* attributes))
 984       )
 985     (when (and (memq 'script attributes)
 986                (setq value (get-char-attribute char 'script)))
 987       (insert (format "(script\t\t%s)%s"
 988                       (mapconcat (function prin1-to-string)
 989                                  value " ")
 990                       line-breaking))
 991       (setq attributes (delq 'script attributes))
 992       )
 993     (dolist (name '(=>ucs =>ucs*))
 994       (when (and (memq name attributes)
 995                  (setq value (get-char-attribute char name)))
 996         (insert (format "(%-18s . #x%04X)\t; %c%s"
 997                         name value (decode-char '=ucs value)
 998                         line-breaking))
 999         (setq attributes (delq name attributes))))
1000     (dolist (name '(=>ucs@gb =>ucs@big5))
1001       (when (and (memq name attributes)
1002                  (setq value (get-char-attribute char name)))
1003         (insert (format "(%-18s . #x%04X)\t; %c%s"
1004                         name value
1005                         (decode-char (intern
1006                                       (concat "="
1007                                               (substring
1008                                                (symbol-name name) 2)))
1009                                      value)
1010                         line-breaking))
1011         (setq attributes (delq name attributes))
1012         ))
1013     ;; (dolist (name '(=>daikanwa))
1014     ;;   (when (and (memq name attributes)
1015     ;;              (setq value (get-char-attribute char name)))
1016     ;;     (insert
1017     ;;      (if (integerp value)
1018     ;;          (format "(%-18s . %05d)\t; %c%s"
1019     ;;                  name value (decode-char '=daikanwa value)
1020     ;;                  line-breaking)
1021     ;;        (format "(%-18s %s)\t; %c%s"
1022     ;;                name
1023     ;;                (mapconcat (function prin1-to-string)
1024     ;;                           value " ")
1025     ;;                (char-representative-of-daikanwa char)
1026     ;;                line-breaking)))
1027     ;;     (setq attributes (delq name attributes))))
1028     (when (and (memq 'general-category attributes)
1029                (setq value (get-char-attribute char 'general-category)))
1030       (insert (format
1031                "(general-category\t%s) ; %s%s"
1032                (mapconcat (lambda (cell)
1033                             (format "%S" cell))
1034                           value " ")
1035                (cond ((rassoc value unidata-normative-category-alist)
1036                       "Normative Category")
1037                      ((rassoc value unidata-informative-category-alist)
1038                       "Informative Category")
1039                      (t
1040                       "Unknown Category"))
1041                line-breaking))
1042       (setq attributes (delq 'general-category attributes))
1043       )
1044     (when (and (memq 'bidi-category attributes)
1045                (setq value (get-char-attribute char 'bidi-category)))
1046       (insert (format "(bidi-category\t. %S)%s"
1047                       value
1048                       line-breaking))
1049       (setq attributes (delq 'bidi-category attributes))
1050       )
1051     (unless (or (not (memq 'mirrored attributes))
1052                 (eq (setq value (get-char-attribute char 'mirrored 'empty))
1053                     'empty))
1054       (insert (format "(mirrored\t\t. %S)%s"
1055                       value
1056                       line-breaking))
1057       (setq attributes (delq 'mirrored attributes))
1058       )
1059     (cond
1060      ((and (memq 'decimal-digit-value attributes)
1061            (setq value (get-char-attribute char 'decimal-digit-value)))
1062       (insert (format "(decimal-digit-value . %S)%s"
1063                       value
1064                       line-breaking))
1065       (setq attributes (delq 'decimal-digit-value attributes))
1066       (when (and (memq 'digit-value attributes)
1067                  (setq value (get-char-attribute char 'digit-value)))
1068         (insert (format "(digit-value\t . %S)%s"
1069                         value
1070                         line-breaking))
1071         (setq attributes (delq 'digit-value attributes))
1072         )
1073       (when (and (memq 'numeric-value attributes)
1074                  (setq value (get-char-attribute char 'numeric-value)))
1075         (insert (format "(numeric-value\t . %S)%s"
1076                         value
1077                         line-breaking))
1078         (setq attributes (delq 'numeric-value attributes))
1079         )
1080       )
1081      (t
1082       (when (and (memq 'digit-value attributes)
1083                  (setq value (get-char-attribute char 'digit-value)))
1084         (insert (format "(digit-value\t. %S)%s"
1085                         value
1086                         line-breaking))
1087         (setq attributes (delq 'digit-value attributes))
1088         )
1089       (when (and (memq 'numeric-value attributes)
1090                  (setq value (get-char-attribute char 'numeric-value)))
1091         (insert (format "(numeric-value\t. %S)%s"
1092                         value
1093                         line-breaking))
1094         (setq attributes (delq 'numeric-value attributes))
1095         )))
1096     (when (and (memq 'iso-10646-comment attributes)
1097                (setq value (get-char-attribute char 'iso-10646-comment)))
1098       (insert (format "(iso-10646-comment\t. %S)%s"
1099                       value
1100                       line-breaking))
1101       (setq attributes (delq 'iso-10646-comment attributes))
1102       )
1103     (when (and (memq 'morohashi-daikanwa attributes)
1104                (setq value (get-char-attribute char 'morohashi-daikanwa)))
1105       (insert (format "(morohashi-daikanwa\t%s)%s"
1106                       (mapconcat (function prin1-to-string) value " ")
1107                       line-breaking))
1108       (setq attributes (delq 'morohashi-daikanwa attributes))
1109       )
1110     (setq radical nil
1111           strokes nil)
1112     (when (and (memq 'ideographic-radical attributes)
1113                (setq value (get-char-attribute char 'ideographic-radical)))
1114       (setq radical value)
1115       (insert (format "(ideographic-radical . %S)\t; %c%s"
1116                       radical
1117                       (ideographic-radical radical)
1118                       line-breaking))
1119       (setq attributes (delq 'ideographic-radical attributes))
1120       )
1121     (when (and (memq 'shuowen-radical attributes)
1122                (setq value (get-char-attribute char 'shuowen-radical)))
1123       (insert (format "(shuowen-radical\t. %S)\t; %c%s"
1124                       value
1125                       (shuowen-radical value)
1126                       line-breaking))
1127       (setq attributes (delq 'shuowen-radical attributes))
1128       )
1129     (let (key)
1130       (dolist (domain
1131                (append
1132                 char-db-feature-domains
1133                 (let (dest domain)
1134                   (dolist (feature (char-attribute-list))
1135                     (setq feature (symbol-name feature))
1136                     (when (string-match
1137                            "\\(radical\\|strokes\\)@\\([^@*]+\\)\\(\\*\\|$\\)"
1138                            feature)
1139                       (setq domain (intern (match-string 2 feature)))
1140                      (unless (memq domain dest)
1141                        (setq dest (cons domain dest)))))
1142                   (sort dest #'string<))))
1143         (setq key (intern (format "%s@%s" 'ideographic-radical domain)))
1144         (when (and (memq key attributes)
1145                    (setq value (get-char-attribute char key)))
1146           (setq radical value)
1147           (insert (format "(%s . %S)\t; %c%s"
1148                           key
1149                           radical
1150                           (ideographic-radical radical)
1151                           line-breaking))
1152           (setq attributes (delq key attributes))
1153           )
1154         (setq key (intern (format "%s@%s" 'ideographic-strokes domain)))
1155         (when (and (memq key attributes)
1156                    (setq value (get-char-attribute char key)))
1157           (setq strokes value)
1158           (insert (format "(%s . %S)%s"
1159                           key
1160                           strokes
1161                           line-breaking))
1162           (setq attributes (delq key attributes))
1163           )
1164         (setq key (intern (format "%s@%s" 'total-strokes domain)))
1165         (when (and (memq key attributes)
1166                    (setq value (get-char-attribute char key)))
1167           (insert (format "(%s       . %S)%s"
1168                           key
1169                           value
1170                           line-breaking))
1171           (setq attributes (delq key attributes))
1172           )
1173         (dolist (feature '(ideographic-radical
1174                            ideographic-strokes
1175                            total-strokes))
1176           (setq key (intern (format "%s@%s*sources" feature domain)))
1177           (when (and (memq key attributes)
1178                      (setq value (get-char-attribute char key)))
1179             (insert (format "(%s%s" key line-breaking))
1180             (dolist (cell value)
1181               (insert (format " %s" cell)))
1182             (insert ")")
1183             (insert line-breaking)
1184             (setq attributes (delq key attributes))
1185             ))
1186         ))
1187     (when (and (memq 'ideographic-strokes attributes)
1188                (setq value (get-char-attribute char 'ideographic-strokes)))
1189       (setq strokes value)
1190       (insert (format "(ideographic-strokes . %S)%s"
1191                       strokes
1192                       line-breaking))
1193       (setq attributes (delq 'ideographic-strokes attributes))
1194       )
1195     (when (and (memq 'kangxi-radical attributes)
1196                (setq value (get-char-attribute char 'kangxi-radical)))
1197       (unless (eq value radical)
1198         (insert (format "(kangxi-radical\t . %S)\t; %c%s"
1199                         value
1200                         (ideographic-radical value)
1201                         line-breaking))
1202         (or radical
1203             (setq radical value)))
1204       (setq attributes (delq 'kangxi-radical attributes))
1205       )
1206     (when (and (memq 'kangxi-strokes attributes)
1207                (setq value (get-char-attribute char 'kangxi-strokes)))
1208       (unless (eq value strokes)
1209         (insert (format "(kangxi-strokes\t . %S)%s"
1210                         value
1211                         line-breaking))
1212         (or strokes
1213             (setq strokes value)))
1214       (setq attributes (delq 'kangxi-strokes attributes))
1215       )
1216     ;; (when (and (memq 'japanese-radical attributes)
1217     ;;            (setq value (get-char-attribute char 'japanese-radical)))
1218     ;;   (unless (eq value radical)
1219     ;;     (insert (format "(japanese-radical\t . %S)\t; %c%s"
1220     ;;                     value
1221     ;;                     (ideographic-radical value)
1222     ;;                     line-breaking))
1223     ;;     (or radical
1224     ;;         (setq radical value)))
1225     ;;   (setq attributes (delq 'japanese-radical attributes))
1226     ;;   )
1227     (when (and (memq 'japanese-strokes attributes)
1228                (setq value (get-char-attribute char 'japanese-strokes)))
1229       (unless (eq value strokes)
1230         (insert (format "(japanese-strokes\t . %S)%s"
1231                         value
1232                         line-breaking))
1233         (or strokes
1234             (setq strokes value)))
1235       (setq attributes (delq 'japanese-strokes attributes))
1236       )
1237     (when (and (memq 'cns-radical attributes)
1238                (setq value (get-char-attribute char 'cns-radical)))
1239       (insert (format "(cns-radical\t . %S)\t; %c%s"
1240                       value
1241                       (ideographic-radical value)
1242                       line-breaking))
1243       (setq attributes (delq 'cns-radical attributes))
1244       )
1245     (when (and (memq 'cns-strokes attributes)
1246                (setq value (get-char-attribute char 'cns-strokes)))
1247       (unless (eq value strokes)
1248         (insert (format "(cns-strokes\t . %S)%s"
1249                         value
1250                         line-breaking))
1251         (or strokes
1252             (setq strokes value)))
1253       (setq attributes (delq 'cns-strokes attributes))
1254       )
1255     ;; (when (and (memq 'ideographic- attributes)
1256     ;;            (setq value (get-char-attribute char 'ideographic-)))
1257     ;;   (insert "(ideographic-       ")
1258     ;;   (setq lbs (concat "\n" (make-string (current-column) ?\ ))
1259     ;;         separator nil)
1260     ;;   (while (consp value)
1261     ;;     (setq cell (car value))
1262     ;;     (if (integerp cell)
1263     ;;         (setq cell (decode-char '=ucs cell)))
1264     ;;     (cond ((characterp cell)
1265     ;;            (if separator
1266     ;;                (insert lbs))
1267     ;;            (if readable
1268     ;;                (insert (format "%S" cell))
1269     ;;              (char-db-insert-char-spec cell readable))
1270     ;;            (setq separator lbs))
1271     ;;           ((consp cell)
1272     ;;            (if separator
1273     ;;                (insert lbs))
1274     ;;            (if (consp (car cell))
1275     ;;                (char-db-insert-char-spec cell readable)
1276     ;;              (char-db-insert-char-reference cell readable))
1277     ;;            (setq separator lbs))
1278     ;;           (t
1279     ;;            (if separator
1280     ;;                (insert separator))
1281     ;;            (insert (prin1-to-string cell))
1282     ;;            (setq separator " ")))
1283     ;;     (setq value (cdr value)))
1284     ;;   (insert ")")
1285     ;;   (insert line-breaking)
1286     ;;   (setq attributes (delq 'ideographic- attributes)))
1287     (when (and (memq 'total-strokes attributes)
1288                (setq value (get-char-attribute char 'total-strokes)))
1289       (insert (format "(total-strokes       . %S)%s"
1290                       value
1291                       line-breaking))
1292       (setq attributes (delq 'total-strokes attributes))
1293       )
1294     ;; (when (and (memq '->ideograph attributes)
1295     ;;            (setq value (get-char-attribute char '->ideograph)))
1296     ;;   (insert (format "(->ideograph\t%s)%s"
1297     ;;                   (mapconcat (lambda (code)
1298     ;;                                (cond ((symbolp code)
1299     ;;                                       (symbol-name code))
1300     ;;                                      ((integerp code)
1301     ;;                                       (format "#x%04X" code))
1302     ;;                                      (t
1303     ;;                                       (format "%s %S"
1304     ;;                                               line-breaking code))))
1305     ;;                              value " ")
1306     ;;                   line-breaking))
1307     ;;   (setq attributes (delq '->ideograph attributes))
1308     ;;   )
1309     ;; (when (and (memq '->decomposition attributes)
1310     ;;            (setq value (get-char-attribute char '->decomposition)))
1311     ;;   (insert (format "(->decomposition\t%s)%s"
1312     ;;                   (mapconcat (lambda (code)
1313     ;;                                (cond ((symbolp code)
1314     ;;                                       (symbol-name code))
1315     ;;                                      ((characterp code)
1316     ;;                                       (if readable
1317     ;;                                           (format "%S" code)
1318     ;;                                         (format "#x%04X"
1319     ;;                                                 (char-int code))
1320     ;;                                         ))
1321     ;;                                      ((integerp code)
1322     ;;                                       (format "#x%04X" code))
1323     ;;                                      (t
1324     ;;                                       (format "%s%S" line-breaking code))))
1325     ;;                              value " ")
1326     ;;                   line-breaking))
1327     ;;   (setq attributes (delq '->decomposition attributes))
1328     ;;   )
1329     (if (equal (get-char-attribute char '->titlecase)
1330                (get-char-attribute char '->uppercase))
1331         (setq attributes (delq '->titlecase attributes)))
1332     (when (and (memq '->mojikyo attributes)
1333                (setq value (get-char-attribute char '->mojikyo)))
1334       (insert (format "(->mojikyo\t\t. %06d)\t; %c%s"
1335                       value (decode-char 'mojikyo value)
1336                       line-breaking))
1337       (setq attributes (delq '->mojikyo attributes))
1338       )
1339     (unless readable
1340       (dolist (ignored '(composition
1341                          ->denotational <-subsumptive ->ucs-unified
1342                          ;; ->ideographic-component-forms
1343                          ))
1344         (setq attributes (delq ignored attributes))))
1345     (while attributes
1346       (setq name (car attributes))
1347       (unless (eq (setq value (get-char-attribute char name 'value-is-empty))
1348                   'value-is-empty)
1349         (cond ((setq ret (find-charset name))
1350                (setq name (charset-name ret))
1351                (when (not (memq name dest-ccss))
1352                  (setq dest-ccss (cons name dest-ccss))
1353                  (char-db-insert-ccs-feature name value line-breaking))
1354                )
1355               ((string-match "^=>ucs@" (symbol-name name))
1356                (insert (format "(%-18s . #x%04X)\t; %c%s"
1357                                name value (decode-char '=ucs value)
1358                                line-breaking))
1359                )
1360               ((eq name 'jisx0208-1978/4X)
1361                (insert (format "(%-18s . #x%04X)%s"
1362                                name value
1363                                line-breaking))
1364                )
1365               ((and
1366                 (not readable)
1367                 (not (eq name '->subsumptive))
1368                 (not (eq name '->uppercase))
1369                 (not (eq name '->lowercase))
1370                 (not (eq name '->titlecase))
1371                 (not (eq name '->canonical))
1372                 (not (eq name '->Bopomofo))
1373                 (not (eq name '->mistakable))
1374                 (not (eq name '->ideographic-variants))
1375                 (null (get-char-attribute
1376                        char (intern (format "%s*sources" name))))
1377                 (not (string-match "\\*sources$" (symbol-name name)))
1378                 (null (get-char-attribute
1379                        char (intern (format "%s*note" name))))
1380                 (not (string-match "\\*note$" (symbol-name name)))
1381                 (or (eq name '<-identical)
1382                     (eq name '<-uppercase)
1383                     (eq name '<-lowercase)
1384                     (eq name '<-titlecase)
1385                     (eq name '<-canonical)
1386                     (eq name '<-ideographic-variants)
1387                     ;; (eq name '<-synonyms)
1388                     (string-match "^<-synonyms" (symbol-name name))
1389                     (eq name '<-mistakable)
1390                     (when (string-match "^->" (symbol-name name))
1391                       (cond
1392                        ((string-match "^->fullwidth" (symbol-name name))
1393                         (not (and (consp value)
1394                                   (characterp (car value))
1395                                   (encode-char
1396                                    (car value) '=ucs 'defined-only)))
1397                         )
1398                        (t)))
1399                     ))
1400                )
1401               ((or (eq name 'ideographic-structure)
1402                    (eq name 'ideographic-combination)
1403                    ;; (eq name 'ideographic-)
1404                    (eq name '=decomposition)
1405                    (char-feature-base-name= '=decomposition name)
1406                    (char-feature-base-name= '=>decomposition name)
1407                    ;; (string-match "^=>*decomposition\\(@[^*]+\\)?$"
1408                    ;;               (symbol-name name))
1409                    (string-match "^\\(->\\|<-\\)[^*]*$" (symbol-name name))
1410                    (string-match "^\\(->\\|<-\\)[^*]*\\*sources$"
1411                                  (symbol-name name))
1412                    )
1413                (char-db-insert-relation-feature char name value
1414                                                 line-breaking
1415                                                 ccss readable))
1416               ;; ((memq name '(ideograph=
1417               ;;               original-ideograph-of
1418               ;;               ancient-ideograph-of
1419               ;;               vulgar-ideograph-of
1420               ;;               wrong-ideograph-of
1421               ;;               ;; simplified-ideograph-of
1422               ;;               ideographic-variants
1423               ;;               ;; ideographic-different-form-of
1424               ;;               ))
1425               ;;  (insert (format "(%-18s%s " name line-breaking))
1426               ;;  (setq lbs (concat "\n" (make-string (current-column) ?\ ))
1427               ;;        separator nil)
1428               ;;  (while (consp value)
1429               ;;    (setq cell (car value))
1430               ;;    (if (and (consp cell)
1431               ;;             (consp (car cell)))
1432               ;;        (progn
1433               ;;          (if separator
1434               ;;              (insert lbs))
1435               ;;          (char-db-insert-alist cell readable)
1436               ;;          (setq separator lbs))
1437               ;;      (if separator
1438               ;;          (insert separator))
1439               ;;      (insert (prin1-to-string cell))
1440               ;;      (setq separator " "))
1441               ;;    (setq value (cdr value)))
1442               ;;  (insert ")")
1443               ;;  (insert line-breaking))
1444               ((consp value)
1445                (insert (format "(%-18s " name))
1446                (setq lbs (concat "\n" (make-string (current-column) ?\ ))
1447                      separator nil)
1448                (while (consp value)
1449                  (setq cell (car value))
1450                  (if (and (consp cell)
1451                           (consp (car cell))
1452                           (setq ret (condition-case nil
1453                                         (find-char cell)
1454                                       (error nil))))
1455                      (progn
1456                        (setq rest cell
1457                              al nil
1458                              cal nil)
1459                        (while rest
1460                          (setq key (car (car rest)))
1461                          (if (find-charset key)
1462                              (setq cal (cons key cal))
1463                            (setq al (cons key al)))
1464                          (setq rest (cdr rest)))
1465                        (if separator
1466                            (insert lbs))
1467                        (insert-char-attributes ret
1468                                                readable
1469                                                al ; cal
1470                                                nil 'for-sub-node)
1471                        (setq separator lbs))
1472                    (setq ret (prin1-to-string cell))
1473                    (if separator
1474                        (if (< (+ (current-column)
1475                                  (length ret)
1476                                  (length separator))
1477                               76)
1478                            (insert separator)
1479                          (insert lbs)))
1480                    (insert ret)
1481                    (setq separator " "))
1482                  (setq value (cdr value)))
1483                (insert ")")
1484                (insert line-breaking))
1485               (t
1486                (insert (format "(%-18s" name))
1487                (setq ret (prin1-to-string value))
1488                (unless (< (+ (current-column)
1489                              (length ret)
1490                              3)
1491                           76)
1492                  (insert line-breaking))
1493                (insert " . " ret ")" line-breaking)
1494                ;; (insert (format "(%-18s . %S)%s"
1495                ;;                 name value
1496                ;;                 line-breaking))
1497                )
1498               ))
1499       (setq attributes (cdr attributes)))
1500     (insert ")")))
1501
1502 (defun insert-char-data (char &optional readable
1503                               attributes)
1504   (save-restriction
1505     (narrow-to-region (point)(point))
1506     (insert "(define-char
1507   '")
1508     (insert-char-attributes char readable attributes)
1509     (insert ")\n")
1510     (goto-char (point-min))
1511     (while (re-search-forward "[ \t]+$" nil t)
1512       (replace-match ""))
1513     ;; from tabify.
1514     (goto-char (point-min))
1515     (while (re-search-forward "[ \t][ \t][ \t]*" nil t)
1516       (let ((column (current-column))
1517             (indent-tabs-mode t))
1518         (delete-region (match-beginning 0) (point))
1519         (indent-to column)))
1520     (goto-char (point-max))
1521     ;; (tabify (point-min)(point-max))
1522     ))
1523
1524 (defun insert-char-data-with-variant (char &optional printable
1525                                            no-ucs-unified
1526                                            script excluded-script)
1527   (insert-char-data char printable)
1528   (let ((variants (char-variants char))
1529         rest
1530         variant vs ret)
1531     (setq variants (sort variants #'<))
1532     (setq rest variants)
1533     (setq variants (cons char variants))
1534     (while rest
1535       (setq variant (car rest))
1536       (unless (get-char-attribute variant '<-subsumptive)
1537         (if (and (or (null script)
1538                      (null (setq vs (get-char-attribute variant 'script)))
1539                      (memq script vs))
1540                  (or (null excluded-script)
1541                      (null (setq vs (get-char-attribute variant 'script)))
1542                      (not (memq excluded-script vs))))
1543             (unless (and no-ucs-unified (get-char-attribute variant '=ucs))
1544               (insert-char-data variant printable)
1545               (if (setq ret (char-variants variant))
1546                   (while ret
1547                     (or (memq (car ret) variants)
1548                         ;; (get-char-attribute (car ret) '<-subsumptive)
1549                         (setq rest (nconc rest (list (car ret)))))
1550                     (setq ret (cdr ret)))))))
1551       (setq rest (cdr rest)))))
1552
1553 (defun insert-char-range-data (min max &optional script excluded-script)
1554   (let ((code min)
1555         char)
1556     (while (<= code max)
1557       (setq char (decode-char '=ucs code))
1558       (if (encode-char char '=ucs 'defined-only)
1559           (insert-char-data-with-variant char nil 'no-ucs-unified
1560                                          script excluded-script))
1561       (setq code (1+ code)))))
1562
1563 (defun write-char-range-data-to-file (min max file
1564                                           &optional script excluded-script)
1565   (let ((coding-system-for-write char-db-file-coding-system))
1566     (with-temp-buffer
1567       (insert (format ";; -*- coding: %s -*-\n"
1568                       char-db-file-coding-system))
1569       (insert-char-range-data min max script excluded-script)
1570       (write-region (point-min)(point-max) file))))
1571
1572 (defvar what-character-original-window-configuration)
1573
1574 ;;;###autoload
1575 (defun what-char-definition (char)
1576   (interactive (list (char-after)))
1577   (let ((buf (get-buffer-create "*Character Description*"))
1578         (the-buf (current-buffer))
1579         (win-conf (current-window-configuration)))
1580     (pop-to-buffer buf)
1581     (make-local-variable 'what-character-original-window-configuration)
1582     (setq what-character-original-window-configuration win-conf)
1583     (setq buffer-read-only nil)
1584     (erase-buffer)
1585     (condition-case err
1586         (progn
1587           (insert-char-data-with-variant char 'printable)
1588           (unless (char-attribute-alist char)
1589             (insert (format ";; = %c\n"
1590                             (let* ((rest (split-char char))
1591                                    (ccs (pop rest))
1592                                    (code (pop rest)))
1593                               (while rest
1594                                 (setq code (logior (lsh code 8)
1595                                                    (pop rest))))
1596                               (decode-char ccs code)))))
1597           ;; (char-db-update-comment)
1598           (set-buffer-modified-p nil)
1599           (view-mode the-buf (lambda (buf)
1600                                (set-window-configuration
1601                                 what-character-original-window-configuration)
1602                                ))
1603           (goto-char (point-min)))
1604       (error (progn
1605                (set-window-configuration
1606                 what-character-original-window-configuration)
1607                (signal (car err) (cdr err)))))))
1608
1609
1610 ;;; @ end
1611 ;;;
1612
1613 (provide 'char-db-util)
1614
1615 ;;; char-db-util.el ends here