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