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