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 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/ib
 189     ==hanyo-denshi/hg
 190     ==hanyo-denshi/jt
 191     ==hanyo-denshi/ks
 192     =gt
 193     =gt-k
 194     =daikanwa
 195     =daikanwa@rev2
 196     =daikanwa@rev1
 197     =daikanwa/+p
 198     ==daikanwa
 199     =cbeta
 200     ideograph-hanziku-1
 201     ideograph-hanziku-2
 202     ideograph-hanziku-3
 203     ideograph-hanziku-4
 204     ideograph-hanziku-5
 205     ideograph-hanziku-6
 206     ideograph-hanziku-7
 207     ideograph-hanziku-8
 208     ideograph-hanziku-9
 209     ideograph-hanziku-10
 210     ideograph-hanziku-11
 211     ideograph-hanziku-12
 212     ;; =>>>adobe-japan1-0
 213     ;; =>>>adobe-japan1-1
 214     ;; =>>>adobe-japan1-2
 215     ;; =>>>adobe-japan1-3
 216     ;; =>>>adobe-japan1-4
 217     ;; =>>>adobe-japan1-5
 218     ;; =>>>adobe-japan1-6
 219     ;; =>>>jis-x0208
 220     ;; =>>>jis-x0213-1
 221     ;; =>>>jis-x0213-2
 222     ;; =>>>hanyo-denshi/ja
 223     ;; =>>>hanyo-denshi/jb
 224     ;; =>>>hanyo-denshi/jc
 225     ;; =>>>hanyo-denshi/ft
 226     ;; =>>>hanyo-denshi/ib
 227     ;; =>>>hanyo-denshi/hg
 228     ;; =>>>hanyo-denshi/jt
 229     ;; =>>>hanyo-denshi/ks
 230     ;; =>>>gt
 231     =>>adobe-japan1-0
 232     =>>adobe-japan1-1
 233     =>>adobe-japan1-2
 234     =>>adobe-japan1-3
 235     =>>adobe-japan1-4
 236     =>>adobe-japan1-5
 237     =>>adobe-japan1-6
 238     =>>jis-x0208
 239     =>>jis-x0213-1
 240     =>>jis-x0213-1@2000
 241     =>>jis-x0213-1@2004
 242     =>>jis-x0213-2
 243     =>>jis-x0208@1978
 244     =>>hanyo-denshi/ft
 245     =>>hanyo-denshi/jt
 246     =>>hanyo-denshi/ks
 247     =>>gt
 248     =>>daikanwa
 249     =>>cbeta
 250     =+>jis-x0208
 251     =+>jis-x0213-1
 252     =+>jis-x0213-2
 253     =+>adobe-japan1-0
 254     =+>adobe-japan1-1
 255     =+>adobe-japan1-2
 256     =+>adobe-japan1-3
 257     =+>adobe-japan1-4
 258     =+>adobe-japan1-5
 259     =+>adobe-japan1-6
 260     =+>jis-x0208@1978
 261     =>jis-x0208
 262     =>jis-x0208@1997
 263     =>jis-x0213-1
 264     =>jis-x0213-1@2000
 265     =>jis-x0213-1@2004
 266     =>jis-x0213-2
 267     ==>ucs@bucs
 268     =>iwds-1
 269     =>ucs@hanyo-denshi
 270     =>ucs@iso
 271     =>ucs@unicode
 272     =>ucs@jis
 273     =>ucs@cns
 274     =>ucs@ks
 275     =+>ucs@iso
 276     =+>ucs@unicode
 277     =+>ucs@jis
 278     =+>ucs@jis/1990
 279     =+>ucs@cns
 280     =+>ucs@ks
 281     =>>ucs@iso
 282     =>>ucs@unicode
 283     =>>ucs@jis
 284     =>>ucs@cns
 285     =>>>ucs@iso
 286     =>>>ucs@unicode
 287     ==ucs@iso
 288     ==ucs@unicode
 289     ;; ==ucs@cns
 290     ==gb2312
 291     ==ks-x1001
 292     ==cns11643-1
 293     ==cns11643-2
 294     ==cns11643-3
 295     ==cns11643-4
 296     ==cns11643-5
 297     ==cns11643-6
 298     ==cns11643-7
 299     ==gt
 300     ==jis-x0208@1990
 301     ;; ==jis-x0208@1983
 302     ==jis-x0208@1978
 303     ==gt-k
 304     =ucs@iso
 305     =ucs@unicode
 306     =ucs@cns
 307     ==big5-cdp
 308     ==cbeta
 309     =>>big5-cdp
 310     =>>gt-k
 311     =+>gt
 312     =>gt
 313     =>mj
 314     =>big5-cdp
 315     =>daikanwa
 316     =>daikanwa/ho
 317     =>cns11643-5
 318     =>cns11643-7
 319     =big5
 320     =big5-eten
 321     =>gt-k
 322     =zinbun-oracle
 323     =>zinbun-oracle
 324     =ruimoku-v6
 325     =>>ruimoku-v6
 326     ==ruimoku-v6
 327     =jef-china3
 328     =>cbeta
 329     =shinjigen
 330     =ucs-var-001
 331     =ucs-var-002
 332     =ucs-var-003
 333     =ucs-var-004
 334     =ucs-var-010
 335     =ucs-itaiji-001
 336     =ucs-itaiji-002
 337     =ucs-itaiji-003
 338     =ucs-itaiji-004
 339     =ucs-itaiji-005
 340     =ucs-itaiji-006
 341     =ucs-itaiji-008
 342     =big5-cdp-var-003
 343     =big5-cdp-var-004
 344     =big5-cdp-var-005
 345     =>ucs@iwds-1
 346     ==ucs@gb))
 347
 348
 349 ;;; @ char-db formatters
 350 ;;;
 351
 352 (defun char-db-make-char-spec (char)
 353   (let (ret char-spec)
 354     (cond ((characterp char)
 355            (cond ((and (setq ret (encode-char char '=ucs 'defined-only))
 356                        (not (and (<= #xE000 ret)(<= ret #xF8FF))))
 357                   (setq char-spec (list (cons '=ucs ret)))
 358                   (cond ((setq ret (get-char-attribute char 'name))
 359                          (setq char-spec (cons (cons 'name ret) char-spec))
 360                          )
 361                         ((setq ret (get-char-attribute char 'name*))
 362                          (setq char-spec (cons (cons 'name* ret) char-spec))
 363                          ))
 364                   )
 365                  ((encode-char char '=mj 'defined-only)
 366                   (setq char-spec nil)
 367                   (dolist (ccs (charset-list))
 368                     (if (and (or (eq ccs '=mj)
 369                                  ;; (eq (charset-property ccs 'iso-ir) 177)
 370                                  (string-match "=ucs@" (symbol-name ccs))
 371                                  )
 372                              (setq ccs (charset-name ccs))
 373                              (null (assq ccs char-spec))
 374                              (setq ret (encode-char char ccs 'defined-only)))
 375                         (setq char-spec (cons (cons ccs ret) char-spec))))
 376                   )
 377                  ((encode-char char '==mj 'defined-only)
 378                   (setq char-spec nil)
 379                   (dolist (ccs (charset-list))
 380                     (if (and (or (eq ccs '==mj)
 381                                  ;; (eq (charset-property ccs 'iso-ir) 177)
 382                                  (string-match "=ucs@" (symbol-name ccs))
 383                                  )
 384                              (setq ccs (charset-name ccs))
 385                              (null (assq ccs char-spec))
 386                              (setq ret (encode-char char ccs 'defined-only)))
 387                         (setq char-spec (cons (cons ccs ret) char-spec))))
 388                   )
 389                  ((encode-char char '=adobe-japan1 'defined-only)
 390                   (setq char-spec nil)
 391                   (dolist (ccs (charset-list))
 392                     (if (and (or (memq ccs
 393                                        '(=adobe-japan1-0
 394                                          =adobe-japan1-1
 395                                          =adobe-japan1-2
 396                                          =adobe-japan1-3
 397                                          =adobe-japan1-4
 398                                          =adobe-japan1-5
 399                                          =adobe-japan1-6
 400                                          ))
 401                                  ;; (eq (charset-property ccs 'iso-ir) 177)
 402                                  (string-match "=ucs@" (symbol-name ccs))
 403                                  )
 404                              (setq ccs (charset-name ccs))
 405                              (null (assq ccs char-spec))
 406                              (setq ret (encode-char char ccs 'defined-only)))
 407                         (setq char-spec (cons (cons ccs ret) char-spec))))
 408                   )
 409                  ((encode-char char '==adobe-japan1 'defined-only)
 410                   (setq char-spec nil)
 411                   (dolist (ccs (charset-list))
 412                     (if (and (or (memq ccs
 413                                        '(==adobe-japan1-0
 414                                          ==adobe-japan1-1
 415                                          ==adobe-japan1-2
 416                                          ==adobe-japan1-3
 417                                          ==adobe-japan1-4
 418                                          ==adobe-japan1-5
 419                                          ==adobe-japan1-6
 420                                          ))
 421                                  ;; (eq (charset-property ccs 'iso-ir) 177)
 422                                  (string-match "=ucs@" (symbol-name ccs))
 423                                  )
 424                              (setq ccs (charset-name ccs))
 425                              (null (assq ccs char-spec))
 426                              (setq ret (encode-char char ccs 'defined-only)))
 427                         (setq char-spec (cons (cons ccs ret) char-spec))))
 428                   )
 429                  ((setq ret
 430                         (catch 'tag
 431                           (let ((rest char-db-coded-charset-priority-list)
 432                                 ccs)
 433                             (while rest
 434                               (setq ccs (charset-name
 435                                          (find-charset (car rest))))
 436                               (if (setq ret
 437                                         (encode-char char ccs
 438                                                      'defined-only))
 439                                   (throw 'tag (cons ccs ret)))
 440                               (setq rest (cdr rest))))))
 441                   (setq char-spec (list ret))
 442                   (dolist (ccs (delq (car ret) (charset-list)))
 443                     (if (and (or (charset-iso-final-char ccs)
 444                                  (memq ccs
 445                                        '(=daikanwa
 446                                          =daikanwa@rev2
 447                                          ;; =gt-k
 448                                          =jis-x0208@1997
 449                                          ))
 450                                  (eq (charset-property ccs 'iso-ir) 177)
 451                                  ;; (string-match "=ucs@" (symbol-name ccs))
 452                                  )
 453                              (setq ccs (charset-name ccs))
 454                              (null (assq ccs char-spec))
 455                              (setq ret (encode-char char ccs 'defined-only)))
 456                         (setq char-spec (cons (cons ccs ret) char-spec))))
 457                   (if (null char-spec)
 458                       (setq char-spec (split-char char)))
 459                   (cond ((setq ret (get-char-attribute char 'name))
 460                          (setq char-spec (cons (cons 'name ret) char-spec))
 461                          )
 462                         ((setq ret (get-char-attribute char 'name*))
 463                          (setq char-spec (cons (cons 'name* ret) char-spec))
 464                          ))
 465                   )
 466                  ((setq ret (get-char-attribute
 467                              char 'ideographic-combination))
 468                   (setq char-spec
 469                         (cons (cons 'ideographic-combination ret)
 470                               char-spec))
 471                   ))
 472            char-spec)
 473           ((consp char)
 474            char))))
 475
 476 (defun char-db-insert-char-spec (char &optional readable column
 477                                       required-features)
 478   (unless column
 479     (setq column (current-column)))
 480   (let (char-spec temp-char)
 481     (setq char-spec (char-db-make-char-spec char))
 482     (unless (or (characterp char) ; char
 483                 (condition-case nil
 484                     (setq char (find-char char-spec))
 485                   (error nil)))
 486       ;; define temporary character
 487       ;;   Current implementation is dirty.
 488       (setq temp-char (define-char (cons '(ideograph-daikanwa . 0)
 489                                          char-spec)))
 490       (remove-char-attribute temp-char 'ideograph-daikanwa)
 491       (setq char temp-char))
 492     (insert-char-attributes char
 493                             readable
 494                             (union (mapcar #'car char-spec)
 495                                    required-features)
 496                             nil 'for-sub-node)
 497     (when temp-char
 498       ;; undefine temporary character
 499       ;;   Current implementation is dirty.
 500       (setq char-spec (char-attribute-alist temp-char))
 501       (while char-spec
 502         (remove-char-attribute temp-char (car (car char-spec)))
 503         (setq char-spec (cdr char-spec))))))
 504
 505 (defun char-db-insert-alist (alist &optional readable column)
 506   (unless column
 507     (setq column (current-column)))
 508   (let ((line-breaking
 509          (concat "\n" (make-string (1+ column) ?\ )))
 510         name value
 511         ret al ; cal
 512         key
 513         lbs cell rest separator)
 514     (insert "(")
 515     (while alist
 516       (setq name (car (car alist))
 517             value (cdr (car alist)))
 518       (cond ((eq name 'char)
 519              (insert "(char . ")
 520              (if (setq ret (condition-case nil
 521                                (find-char value)
 522                              (error nil)))
 523                  (progn
 524                    (setq al nil
 525                          ;; cal nil
 526                          )
 527                    (while value
 528                      (setq key (car (car value)))
 529                      ;; (if (find-charset key)
 530                      ;;     (setq cal (cons key cal))
 531                      (setq al (cons key al))
 532                      ;; )
 533                      (setq value (cdr value)))
 534                    (insert-char-attributes ret
 535                                            readable
 536                                            (or al 'none) ; cal
 537                                            nil 'for-sub-node))
 538                (insert (prin1-to-string value)))
 539              (insert ")")
 540              (insert line-breaking))
 541             ((consp value)
 542              (insert (format "(%-18s " name))
 543              (setq lbs (concat "\n" (make-string (current-column) ?\ )))
 544              (while (consp value)
 545                (setq cell (car value))
 546                (if (and (consp cell)
 547                         (consp (car cell))
 548                         (setq ret (condition-case nil
 549                                       (find-char cell)
 550                                     (error nil)))
 551                         )
 552                    (progn
 553                      (setq rest cell
 554                            al nil
 555                            ;; cal nil
 556                            )
 557                      (while rest
 558                        (setq key (car (car rest)))
 559                        ;; (if (find-charset key)
 560                        ;;     (setq cal (cons key cal))
 561                        (setq al (cons key al))
 562                        ;; )
 563                        (setq rest (cdr rest)))
 564                      (if separator
 565                          (insert lbs))
 566                      (insert-char-attributes ret
 567                                              readable
 568                                              al ; cal
 569                                              nil 'for-sub-node)
 570                      (setq separator lbs))
 571                  (if separator
 572                      (insert separator))
 573                  (insert (prin1-to-string cell))
 574                  (setq separator " "))
 575                (setq value (cdr value)))
 576              (insert ")")
 577              (insert line-breaking))
 578             (t
 579              (insert (format "(%-18s . %S)%s"
 580                              name value
 581                              line-breaking))))
 582       (setq alist (cdr alist))))
 583   (insert ")"))
 584
 585 (defun char-db-insert-char-reference (plist &optional readable column)
 586   (unless column
 587     (setq column (current-column)))
 588   (let ((line-breaking
 589          (concat "\n" (make-string (1+ column) ?\ )))
 590         (separator "")
 591         name value)
 592     (insert "(")
 593     (while plist
 594       (setq name (pop plist))
 595       (setq value (pop plist))
 596       (cond ((eq name :char)
 597              (insert separator)
 598              (insert ":char\t")
 599              (cond ((numberp value)
 600                     (setq value (decode-char '=ucs value)))
 601                    ;; ((consp value)
 602                    ;;  (setq value (or (find-char value)
 603                    ;;                  value)))
 604                    )
 605              (char-db-insert-char-spec value readable)
 606              (insert line-breaking)
 607              (setq separator ""))
 608             ((eq name :radical)
 609              (insert (format "%s%s\t%d ; %c%s"
 610                              separator
 611                              name value
 612                              (ideographic-radical value)
 613                              line-breaking))
 614              (setq separator ""))
 615             (t
 616              (insert (format "%s%s\t%S" separator name value))
 617              (setq separator line-breaking)))
 618       ))
 619   (insert ")"))
 620
 621 (defun char-db-decode-isolated-char (ccs code-point)
 622   (let (ret)
 623     (setq ret
 624           (cond ((eq ccs 'arabic-iso8859-6)
 625                  (decode-char ccs code-point))
 626                 ;; ((eq ccs '=gt)
 627                 ;;  (decode-builtin-char '==gt code-point))
 628                 ((and (memq ccs '(=gt-pj-1
 629                                   =gt-pj-2
 630                                   =gt-pj-3
 631                                   =gt-pj-4
 632                                   =gt-pj-5
 633                                   =gt-pj-6
 634                                   =gt-pj-7
 635                                   =gt-pj-8
 636                                   =gt-pj-9
 637                                   =gt-pj-10
 638                                   =gt-pj-11))
 639                       (setq ret (decode-char ccs code-point))
 640                       (setq ret (encode-char ret '=gt 'defined-only)))
 641                  (decode-builtin-char '=gt ret))
 642                 (t
 643                  (decode-builtin-char ccs code-point))))
 644     (cond ((null ret)
 645            (or (decode-char ccs code-point)
 646                (define-char (list (cons ccs code-point)))))
 647           ((and (<= 0 (char-int ret))
 648                 (<= (char-int ret) #x1F))
 649            (decode-char '=ucs (+ #x2400 (char-int ret))))
 650           ((= (char-int ret) #x7F)
 651            ?\u2421)
 652           (t ret))))
 653
 654 (defvar char-db-convert-obsolete-format t)
 655
 656 (defun char-db-insert-ccs-feature (name value line-breaking)
 657   (cond
 658    ((integerp value)
 659     (insert
 660      (format
 661       (cond
 662        ((memq name '(=>iwds-1
 663                      ==shinjigen
 664                      =shinjigen
 665                      =shinjigen@1ed ==shinjigen@1ed
 666                      =shinjigen@rev ==shinjigen@rev
 667                      =shinjigen/+p@rev ==shinjigen/+p@rev
 668                      ===daikanwa/ho ==daikanwa/ho
 669                      =daikanwa/ho =>>daikanwa/ho =>daikanwa/ho))
 670         "(%-18s .  %04d)\t; %c")
 671        ((eq name '=shinjigen@1ed/24pr)
 672         "(%-18s . %04d)\t; %c")
 673        ((or
 674          (memq name
 675                '(===daikanwa
 676                  ==daikanwa =daikanwa =>>daikanwa =>daikanwa
 677                  =daikanwa@rev1 =daikanwa@rev2
 678                  =daikanwa/+p ==daikanwa/+p ===daikanwa/+p
 679                  =>>daikanwa/+p
 680                  =daikanwa/+2p =>>daikanwa/+2p
 681                  =gt ==gt ===gt
 682                  =>>gt =+>gt =>gt
 683                  =gt-k ==gt-k ===gt-k
 684                  =>>gt-k =>gt-k
 685                  =adobe-japan1-0 ==adobe-japan1-0 ===adobe-japan1-0
 686                  =adobe-japan1-1 ==adobe-japan1-1 ===adobe-japan1-1
 687                  =adobe-japan1-2 ==adobe-japan1-2 ===adobe-japan1-2
 688                  =adobe-japan1-3 ==adobe-japan1-3 ===adobe-japan1-3
 689                  =adobe-japan1-4 ==adobe-japan1-4 ===adobe-japan1-4
 690                  =adobe-japan1-5 ==adobe-japan1-5 ===adobe-japan1-5
 691                  =adobe-japan1-6 ==adobe-japan1-6 ===adobe-japan1-6
 692                  =>>adobe-japan1-0 =+>adobe-japan1-0
 693                  =>>adobe-japan1-1 =+>adobe-japan1-1
 694                  =>>adobe-japan1-2 =+>adobe-japan1-2
 695                  =>>adobe-japan1-3 =+>adobe-japan1-3
 696                  =>>adobe-japan1-4 =+>adobe-japan1-4
 697                  =>>adobe-japan1-5 =+>adobe-japan1-5
 698                  =>>adobe-japan1-6 =+>adobe-japan1-6
 699                  =>cbeta =cbeta =>>cbeta ==cbeta ===cbeta
 700                  =zinbun-oracle =>zinbun-oracle
 701                  ===hng-jou ===hng-keg ===hng-dng ===hng-mam
 702                  ===hng-drt ===hng-kgk ===hng-myz ===hng-kda
 703                  ===hng-khi ===hng-khm ===hng-hok ===hng-kyd ===hng-sok
 704                  ===hng-yhk ===hng-kak ===hng-kar ===hng-kae
 705                  ===hng-sys ===hng-tsu ===hng-tzj
 706                  ===hng-hos ===hng-nak ===hng-jhk
 707                  ===hng-hod ===hng-gok ===hng-ink ===hng-nto
 708                  ===hng-nkm ===hng-k24 ===hng-nkk
 709                  ===hng-kcc ===hng-kcj ===hng-kbk ===hng-sik
 710                  ===hng-skk ===hng-kyu ===hng-ksk ===hng-wan
 711                  ===hng-okd ===hng-wad ===hng-kmi
 712                  ===hng-zkd ===hng-doh ===hng-jyu
 713                  ===hng-tzs ===hng-kss ===hng-kyo
 714                  ===hng-smk))
 715          ;; (string-match "^=adobe-" (symbol-name name))
 716          )
 717         "(%-18s . %05d)\t; %c")
 718        ((memq name '(=hanyo-denshi/ks
 719                      ==hanyo-denshi/ks ===hanyo-denshi/ks
 720                      =>>hanyo-denshi/ks
 721                      =koseki ==koseki
 722                      =mj ==mj ===mj =>>mj =>mj
 723                      =zihai mojikyo))
 724         "(%-18s . %06d)\t; %c")
 725        ((memq name '(=hanyo-denshi/tk ==hanyo-denshi/tk))
 726         "(%-18s . %08d)\t; %c")
 727        ((>= (charset-dimension name) 2)
 728         "(%-18s . #x%04X)\t; %c")
 729        (t
 730         "(%-18s . #x%02X)\t; %c"))
 731       name
 732       (if (= (charset-iso-graphic-plane name) 1)
 733           (logior value
 734                   (cond ((= (charset-dimension name) 1)
 735                          #x80)
 736                         ((= (charset-dimension name) 2)
 737                          #x8080)
 738                         ((= (charset-dimension name) 3)
 739                          #x808080)
 740                         (t 0)))
 741         value)
 742       (char-db-decode-isolated-char name value)))
 743     (if (and (= (charset-chars name) 94)
 744              (= (charset-dimension name) 2))
 745         (insert (format " [%02d-%02d]"
 746                         (- (lsh value -8) 32)
 747                         (- (logand value 255) 32))))
 748     )
 749    (t
 750     (insert (format "(%-18s . %s)" name value))
 751     ))
 752   (insert line-breaking))
 753
 754 (defun char-db-insert-relation-feature (char name value line-breaking
 755                                              ccss readable)
 756   (insert (format "(%-18s%s " name line-breaking))
 757   (let ((lbs (concat "\n" (make-string (current-column) ?\ )))
 758         separator cell sources required-features
 759         ret)
 760     (while (consp value)
 761       (setq cell (car value))
 762       (if (integerp cell)
 763           (setq cell (decode-char '=ucs cell)))
 764       (cond
 765        ((eq name '->subsumptive)
 766         (when (or (not (some (lambda (atr)
 767                                (get-char-attribute cell atr))
 768                              char-db-ignored-attributes))
 769                   (some (lambda (ccs)
 770                           (encode-char cell ccs 'defined-only))
 771                         ccss))
 772           (if separator
 773               (insert lbs))
 774           (let ((char-db-ignored-attributes
 775                  (cons '<-subsumptive
 776                        char-db-ignored-attributes)))
 777             (insert-char-attributes cell readable nil nil 'for-sub-node))
 778           (setq separator lbs))
 779         )
 780        ((characterp cell)
 781         (setq sources
 782               (get-char-attribute
 783                char (intern (format "%s*sources" name))))
 784         (setq required-features nil)
 785         (dolist (source sources)
 786           (cond
 787            ((memq source '(JP
 788                            JP/Jouyou
 789                            shinjigen shinjigen@1ed shinjigen@rev))
 790             (setq required-features
 791                   (union required-features
 792                          '(=jis-x0208
 793                            =jis-x0208@1990
 794                            =jis-x0213-1@2000
 795                            =jis-x0213-1@2004
 796                            =jis-x0213-2
 797                            =jis-x0212
 798                            =jis-x0208@1983
 799                            =jis-x0208@1978
 800                            =shinjigen
 801                            =shinjigen@1ed
 802                            =shinjigen@rev
 803                            =shinjigen/+p@rev))))
 804            ((eq source 'CN)
 805             (setq required-features
 806                   (union required-features
 807                          '(=gb2312
 808                            =gb12345
 809                            =iso-ir165)))))
 810           (cond
 811            ((find-charset (setq ret (intern (format "=%s" source))))
 812             (setq required-features
 813                   (cons ret required-features)))
 814            (t (setq required-features
 815                     (cons source required-features)))))
 816         (cond ((string-match "@JP" (symbol-name name))
 817                (setq required-features
 818                      (union required-features
 819                             '(=jis-x0208
 820                               =jis-x0208@1990
 821                               =jis-x0213-1-2000
 822                               =jis-x0213-2-2000
 823                               =jis-x0212
 824                               =jis-x0208@1983
 825                               =jis-x0208@1978))))
 826               ((string-match "@CN" (symbol-name name))
 827                (setq required-features
 828                      (union required-features
 829                             '(=gb2312
 830                               =gb12345
 831                               =iso-ir165)))))
 832         (if separator
 833             (insert lbs))
 834         (if readable
 835             (insert (format "%S" cell))
 836           (char-db-insert-char-spec cell readable
 837                                     nil
 838                                     required-features))
 839         (setq separator lbs))
 840        ((consp cell)
 841         (if separator
 842             (insert lbs))
 843         (if (consp (car cell))
 844             (char-db-insert-char-spec cell readable)
 845           (char-db-insert-char-reference cell readable))
 846         (setq separator lbs))
 847        (t
 848         (if separator
 849             (insert separator))
 850         (insert (prin1-to-string cell))
 851         (setq separator " ")))
 852       (setq value (cdr value)))
 853     (insert ")")
 854     (insert line-breaking)))
 855
 856 (defun insert-char-attributes (char &optional readable attributes column
 857                                     for-sub-node)
 858   (unless column
 859     (setq column (current-column)))
 860   (let (name value ; has-long-ccs-name
 861         rest
 862         radical strokes
 863         (line-breaking
 864          (concat "\n" (make-string (1+ column) ?\ )))
 865         lbs cell separator ret
 866         key al cal
 867         dest-ccss ; sources required-features
 868         ccss)
 869     (let (atr-d)
 870       (setq attributes
 871             (sort (if attributes
 872                       (if (consp attributes)
 873                           (progn
 874                             (dolist (name attributes)
 875                               (unless (memq name char-db-ignored-attributes)
 876                                 (if (find-charset name)
 877                                     (push name ccss))
 878                                 (push name atr-d)))
 879                             atr-d))
 880                     (dolist (name (char-attribute-list))
 881                       (unless (memq name char-db-ignored-attributes)
 882                         (if (find-charset name)
 883                             (push name ccss))
 884                         (push name atr-d)))
 885                     atr-d)
 886                   #'char-attribute-name<)))
 887     (insert "(")
 888     (when (memq '<-subsumptive attributes)
 889       (when (or readable (not for-sub-node))
 890         (when (setq value (get-char-attribute char '<-subsumptive))
 891           (char-db-insert-relation-feature char '<-subsumptive value
 892                                            line-breaking
 893                                            ccss readable)))
 894       (setq attributes (delq '<-subsumptive attributes)))
 895     (when (and (memq '<-denotational attributes)
 896                (setq value (get-char-attribute char '<-denotational)))
 897       (char-db-insert-relation-feature char '<-denotational value
 898                                        line-breaking
 899                                        ccss readable)
 900       (setq attributes (delq '<-denotational attributes)))
 901     (when (and (memq '<-denotational@component attributes)
 902                (setq value (get-char-attribute char '<-denotational@component)))
 903       (char-db-insert-relation-feature char '<-denotational@component value
 904                                        line-breaking
 905                                        ccss readable)
 906       (setq attributes (delq '<-denotational@component attributes)))
 907     (when (and (memq 'name attributes)
 908                (setq value (get-char-attribute char 'name)))
 909       (insert (format
 910                (if (> (+ (current-column) (length value)) 48)
 911                    "(name . %S)%s"
 912                  "(name               . %S)%s")
 913                value line-breaking))
 914       (setq attributes (delq 'name attributes))
 915       )
 916     (when (and (memq 'name* attributes)
 917                (setq value (get-char-attribute char 'name*)))
 918       (insert (format
 919                (if (> (+ (current-column) (length value)) 48)
 920                    "(name* . %S)%s"
 921                  "(name*              . %S)%s")
 922                value line-breaking))
 923       (setq attributes (delq 'name* attributes))
 924       )
 925     (when (and (memq 'script attributes)
 926                (setq value (get-char-attribute char 'script)))
 927       (insert (format "(script\t\t%s)%s"
 928                       (mapconcat (function prin1-to-string)
 929                                  value " ")
 930                       line-breaking))
 931       (setq attributes (delq 'script attributes))
 932       )
 933     (dolist (name '(=>ucs =>ucs*))
 934       (when (and (memq name attributes)
 935                  (setq value (get-char-attribute char name)))
 936         (insert (format "(%-18s . #x%04X)\t; %c%s"
 937                         name value (decode-char '=ucs value)
 938                         line-breaking))
 939         (setq attributes (delq name attributes))))
 940     (dolist (name '(=>ucs@gb =>ucs@big5))
 941       (when (and (memq name attributes)
 942                  (setq value (get-char-attribute char name)))
 943         (insert (format "(%-18s . #x%04X)\t; %c%s"
 944                         name value
 945                         (decode-char (intern
 946                                       (concat "="
 947                                               (substring
 948                                                (symbol-name name) 2)))
 949                                      value)
 950                         line-breaking))
 951         (setq attributes (delq name attributes))
 952         ))
 953     ;; (dolist (name '(=>daikanwa))
 954     ;;   (when (and (memq name attributes)
 955     ;;              (setq value (get-char-attribute char name)))
 956     ;;     (insert
 957     ;;      (if (integerp value)
 958     ;;          (format "(%-18s . %05d)\t; %c%s"
 959     ;;                  name value (decode-char '=daikanwa value)
 960     ;;                  line-breaking)
 961     ;;        (format "(%-18s %s)\t; %c%s"
 962     ;;                name
 963     ;;                (mapconcat (function prin1-to-string)
 964     ;;                           value " ")
 965     ;;                (char-representative-of-daikanwa char)
 966     ;;                line-breaking)))
 967     ;;     (setq attributes (delq name attributes))))
 968     (when (and (memq 'general-category attributes)
 969                (setq value (get-char-attribute char 'general-category)))
 970       (insert (format
 971                "(general-category\t%s) ; %s%s"
 972                (mapconcat (lambda (cell)
 973                             (format "%S" cell))
 974                           value " ")
 975                (cond ((rassoc value unidata-normative-category-alist)
 976                       "Normative Category")
 977                      ((rassoc value unidata-informative-category-alist)
 978                       "Informative Category")
 979                      (t
 980                       "Unknown Category"))
 981                line-breaking))
 982       (setq attributes (delq 'general-category attributes))
 983       )
 984     (when (and (memq 'bidi-category attributes)
 985                (setq value (get-char-attribute char 'bidi-category)))
 986       (insert (format "(bidi-category\t. %S)%s"
 987                       value
 988                       line-breaking))
 989       (setq attributes (delq 'bidi-category attributes))
 990       )
 991     (unless (or (not (memq 'mirrored attributes))
 992                 (eq (setq value (get-char-attribute char 'mirrored 'empty))
 993                     'empty))
 994       (insert (format "(mirrored\t\t. %S)%s"
 995                       value
 996                       line-breaking))
 997       (setq attributes (delq 'mirrored attributes))
 998       )
 999     (cond
1000      ((and (memq 'decimal-digit-value attributes)
1001            (setq value (get-char-attribute char 'decimal-digit-value)))
1002       (insert (format "(decimal-digit-value . %S)%s"
1003                       value
1004                       line-breaking))
1005       (setq attributes (delq 'decimal-digit-value attributes))
1006       (when (and (memq 'digit-value attributes)
1007                  (setq value (get-char-attribute char 'digit-value)))
1008         (insert (format "(digit-value\t . %S)%s"
1009                         value
1010                         line-breaking))
1011         (setq attributes (delq 'digit-value attributes))
1012         )
1013       (when (and (memq 'numeric-value attributes)
1014                  (setq value (get-char-attribute char 'numeric-value)))
1015         (insert (format "(numeric-value\t . %S)%s"
1016                         value
1017                         line-breaking))
1018         (setq attributes (delq 'numeric-value attributes))
1019         )
1020       )
1021      (t
1022       (when (and (memq 'digit-value attributes)
1023                  (setq value (get-char-attribute char 'digit-value)))
1024         (insert (format "(digit-value\t. %S)%s"
1025                         value
1026                         line-breaking))
1027         (setq attributes (delq 'digit-value attributes))
1028         )
1029       (when (and (memq 'numeric-value attributes)
1030                  (setq value (get-char-attribute char 'numeric-value)))
1031         (insert (format "(numeric-value\t. %S)%s"
1032                         value
1033                         line-breaking))
1034         (setq attributes (delq 'numeric-value attributes))
1035         )))
1036     (when (and (memq 'iso-10646-comment attributes)
1037                (setq value (get-char-attribute char 'iso-10646-comment)))
1038       (insert (format "(iso-10646-comment\t. %S)%s"
1039                       value
1040                       line-breaking))
1041       (setq attributes (delq 'iso-10646-comment attributes))
1042       )
1043     (when (and (memq 'morohashi-daikanwa attributes)
1044                (setq value (get-char-attribute char 'morohashi-daikanwa)))
1045       (insert (format "(morohashi-daikanwa\t%s)%s"
1046                       (mapconcat (function prin1-to-string) value " ")
1047                       line-breaking))
1048       (setq attributes (delq 'morohashi-daikanwa attributes))
1049       )
1050     (setq radical nil
1051           strokes nil)
1052     (when (and (memq 'ideographic-radical attributes)
1053                (setq value (get-char-attribute char 'ideographic-radical)))
1054       (setq radical value)
1055       (insert (format "(ideographic-radical . %S)\t; %c%s"
1056                       radical
1057                       (ideographic-radical radical)
1058                       line-breaking))
1059       (setq attributes (delq 'ideographic-radical attributes))
1060       )
1061     (when (and (memq 'shuowen-radical attributes)
1062                (setq value (get-char-attribute char 'shuowen-radical)))
1063       (insert (format "(shuowen-radical\t. %S)\t; %c%s"
1064                       value
1065                       (shuowen-radical value)
1066                       line-breaking))
1067       (setq attributes (delq 'shuowen-radical attributes))
1068       )
1069     (let (key)
1070       (dolist (domain
1071                (append
1072                 char-db-feature-domains
1073                 (let (dest domain)
1074                   (dolist (feature (char-attribute-list))
1075                     (setq feature (symbol-name feature))
1076                     (when (string-match
1077                            "\\(radical\\|strokes\\)@\\([^@*]+\\)\\(\\*\\|$\\)"
1078                            feature)
1079                       (setq domain (intern (match-string 2 feature)))
1080                      (unless (memq domain dest)
1081                        (setq dest (cons domain dest)))))
1082                   (sort dest #'string<))))
1083         (setq key (intern (format "%s@%s" 'ideographic-radical domain)))
1084         (when (and (memq key attributes)
1085                    (setq value (get-char-attribute char key)))
1086           (setq radical value)
1087           (insert (format "(%s . %S)\t; %c%s"
1088                           key
1089                           radical
1090                           (ideographic-radical radical)
1091                           line-breaking))
1092           (setq attributes (delq key attributes))
1093           )
1094         (setq key (intern (format "%s@%s" 'ideographic-strokes domain)))
1095         (when (and (memq key attributes)
1096                    (setq value (get-char-attribute char key)))
1097           (setq strokes value)
1098           (insert (format "(%s . %S)%s"
1099                           key
1100                           strokes
1101                           line-breaking))
1102           (setq attributes (delq key attributes))
1103           )
1104         (setq key (intern (format "%s@%s" 'total-strokes domain)))
1105         (when (and (memq key attributes)
1106                    (setq value (get-char-attribute char key)))
1107           (insert (format "(%s       . %S)%s"
1108                           key
1109                           value
1110                           line-breaking))
1111           (setq attributes (delq key attributes))
1112           )
1113         (dolist (feature '(ideographic-radical
1114                            ideographic-strokes
1115                            total-strokes))
1116           (setq key (intern (format "%s@%s*sources" feature domain)))
1117           (when (and (memq key attributes)
1118                      (setq value (get-char-attribute char key)))
1119             (insert (format "(%s%s" key line-breaking))
1120             (dolist (cell value)
1121               (insert (format " %s" cell)))
1122             (insert ")")
1123             (insert line-breaking)
1124             (setq attributes (delq key attributes))
1125             ))
1126         ))
1127     (when (and (memq 'ideographic-strokes attributes)
1128                (setq value (get-char-attribute char 'ideographic-strokes)))
1129       (setq strokes value)
1130       (insert (format "(ideographic-strokes . %S)%s"
1131                       strokes
1132                       line-breaking))
1133       (setq attributes (delq 'ideographic-strokes attributes))
1134       )
1135     (when (and (memq 'kangxi-radical attributes)
1136                (setq value (get-char-attribute char 'kangxi-radical)))
1137       (unless (eq value radical)
1138         (insert (format "(kangxi-radical\t . %S)\t; %c%s"
1139                         value
1140                         (ideographic-radical value)
1141                         line-breaking))
1142         (or radical
1143             (setq radical value)))
1144       (setq attributes (delq 'kangxi-radical attributes))
1145       )
1146     (when (and (memq 'kangxi-strokes attributes)
1147                (setq value (get-char-attribute char 'kangxi-strokes)))
1148       (unless (eq value strokes)
1149         (insert (format "(kangxi-strokes\t . %S)%s"
1150                         value
1151                         line-breaking))
1152         (or strokes
1153             (setq strokes value)))
1154       (setq attributes (delq 'kangxi-strokes attributes))
1155       )
1156     ;; (when (and (memq 'japanese-radical attributes)
1157     ;;            (setq value (get-char-attribute char 'japanese-radical)))
1158     ;;   (unless (eq value radical)
1159     ;;     (insert (format "(japanese-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 'japanese-radical attributes))
1166     ;;   )
1167     (when (and (memq 'japanese-strokes attributes)
1168                (setq value (get-char-attribute char 'japanese-strokes)))
1169       (unless (eq value strokes)
1170         (insert (format "(japanese-strokes\t . %S)%s"
1171                         value
1172                         line-breaking))
1173         (or strokes
1174             (setq strokes value)))
1175       (setq attributes (delq 'japanese-strokes attributes))
1176       )
1177     (when (and (memq 'cns-radical attributes)
1178                (setq value (get-char-attribute char 'cns-radical)))
1179       (insert (format "(cns-radical\t . %S)\t; %c%s"
1180                       value
1181                       (ideographic-radical value)
1182                       line-breaking))
1183       (setq attributes (delq 'cns-radical attributes))
1184       )
1185     (when (and (memq 'cns-strokes attributes)
1186                (setq value (get-char-attribute char 'cns-strokes)))
1187       (unless (eq value strokes)
1188         (insert (format "(cns-strokes\t . %S)%s"
1189                         value
1190                         line-breaking))
1191         (or strokes
1192             (setq strokes value)))
1193       (setq attributes (delq 'cns-strokes attributes))
1194       )
1195     ;; (when (and (memq 'shinjigen-1-radical attributes)
1196     ;;            (setq value (get-char-attribute char 'shinjigen-1-radical)))
1197     ;;   (unless (eq value radical)
1198     ;;     (insert (format "(shinjigen-1-radical . %S)\t; %c%s"
1199     ;;                     value
1200     ;;                     (ideographic-radical value)
1201     ;;                     line-breaking))
1202     ;;     (or radical
1203     ;;         (setq radical value)))
1204     ;;   (setq attributes (delq 'shinjigen-1-radical attributes))
1205     ;;   )
1206     ;; (when (and (memq 'ideographic- attributes)
1207     ;;            (setq value (get-char-attribute char 'ideographic-)))
1208     ;;   (insert "(ideographic-       ")
1209     ;;   (setq lbs (concat "\n" (make-string (current-column) ?\ ))
1210     ;;         separator nil)
1211     ;;   (while (consp value)
1212     ;;     (setq cell (car value))
1213     ;;     (if (integerp cell)
1214     ;;         (setq cell (decode-char '=ucs cell)))
1215     ;;     (cond ((characterp cell)
1216     ;;            (if separator
1217     ;;                (insert lbs))
1218     ;;            (if readable
1219     ;;                (insert (format "%S" cell))
1220     ;;              (char-db-insert-char-spec cell readable))
1221     ;;            (setq separator lbs))
1222     ;;           ((consp cell)
1223     ;;            (if separator
1224     ;;                (insert lbs))
1225     ;;            (if (consp (car cell))
1226     ;;                (char-db-insert-char-spec cell readable)
1227     ;;              (char-db-insert-char-reference cell readable))
1228     ;;            (setq separator lbs))
1229     ;;           (t
1230     ;;            (if separator
1231     ;;                (insert separator))
1232     ;;            (insert (prin1-to-string cell))
1233     ;;            (setq separator " ")))
1234     ;;     (setq value (cdr value)))
1235     ;;   (insert ")")
1236     ;;   (insert line-breaking)
1237     ;;   (setq attributes (delq 'ideographic- attributes)))
1238     (when (and (memq 'total-strokes attributes)
1239                (setq value (get-char-attribute char 'total-strokes)))
1240       (insert (format "(total-strokes       . %S)%s"
1241                       value
1242                       line-breaking))
1243       (setq attributes (delq 'total-strokes attributes))
1244       )
1245     ;; (when (and (memq '->ideograph attributes)
1246     ;;            (setq value (get-char-attribute char '->ideograph)))
1247     ;;   (insert (format "(->ideograph\t%s)%s"
1248     ;;                   (mapconcat (lambda (code)
1249     ;;                                (cond ((symbolp code)
1250     ;;                                       (symbol-name code))
1251     ;;                                      ((integerp code)
1252     ;;                                       (format "#x%04X" code))
1253     ;;                                      (t
1254     ;;                                       (format "%s %S"
1255     ;;                                               line-breaking code))))
1256     ;;                              value " ")
1257     ;;                   line-breaking))
1258     ;;   (setq attributes (delq '->ideograph attributes))
1259     ;;   )
1260     ;; (when (and (memq '->decomposition attributes)
1261     ;;            (setq value (get-char-attribute char '->decomposition)))
1262     ;;   (insert (format "(->decomposition\t%s)%s"
1263     ;;                   (mapconcat (lambda (code)
1264     ;;                                (cond ((symbolp code)
1265     ;;                                       (symbol-name code))
1266     ;;                                      ((characterp code)
1267     ;;                                       (if readable
1268     ;;                                           (format "%S" code)
1269     ;;                                         (format "#x%04X"
1270     ;;                                                 (char-int code))
1271     ;;                                         ))
1272     ;;                                      ((integerp code)
1273     ;;                                       (format "#x%04X" code))
1274     ;;                                      (t
1275     ;;                                       (format "%s%S" line-breaking code))))
1276     ;;                              value " ")
1277     ;;                   line-breaking))
1278     ;;   (setq attributes (delq '->decomposition attributes))
1279     ;;   )
1280     (if (equal (get-char-attribute char '->titlecase)
1281                (get-char-attribute char '->uppercase))
1282         (setq attributes (delq '->titlecase attributes)))
1283     (when (and (memq '->mojikyo attributes)
1284                (setq value (get-char-attribute char '->mojikyo)))
1285       (insert (format "(->mojikyo\t\t. %06d)\t; %c%s"
1286                       value (decode-char 'mojikyo value)
1287                       line-breaking))
1288       (setq attributes (delq '->mojikyo attributes))
1289       )
1290     ;; (when (and (memq 'hanyu-dazidian-vol attributes)
1291     ;;            (setq value (get-char-attribute char 'hanyu-dazidian-vol)))
1292     ;;   (insert (format "(hanyu-dazidian-vol  . %d)%s"
1293     ;;                   value line-breaking))
1294     ;;   (setq attributes (delq 'hanyu-dazidian-vol attributes))
1295     ;;   )
1296     ;; (when (and (memq 'hanyu-dazidian-page attributes)
1297     ;;            (setq value (get-char-attribute char 'hanyu-dazidian-page)))
1298     ;;   (insert (format "(hanyu-dazidian-page . %d)%s"
1299     ;;                   value line-breaking))
1300     ;;   (setq attributes (delq 'hanyu-dazidian-page attributes))
1301     ;;   )
1302     ;; (when (and (memq 'hanyu-dazidian-char attributes)
1303     ;;            (setq value (get-char-attribute char 'hanyu-dazidian-char)))
1304     ;;   (insert (format "(hanyu-dazidian-char . %d)%s"
1305     ;;                   value line-breaking))
1306     ;;   (setq attributes (delq 'hanyu-dazidian-char attributes))
1307     ;;   )
1308     (unless readable
1309       (dolist (ignored '(composition
1310                          ->denotational <-subsumptive ->ucs-unified
1311                          ->ideographic-component-forms))
1312         (setq attributes (delq ignored attributes))))
1313     (while attributes
1314       (setq name (car attributes))
1315       (unless (eq (setq value (get-char-attribute char name 'value-is-empty))
1316                   'value-is-empty)
1317         (cond ((setq ret (find-charset name))
1318                (setq name (charset-name ret))
1319                (when (not (memq name dest-ccss))
1320                  (setq dest-ccss (cons name dest-ccss))
1321                  (char-db-insert-ccs-feature name value line-breaking))
1322                )
1323               ((string-match "^=>ucs@" (symbol-name name))
1324                (insert (format "(%-18s . #x%04X)\t; %c%s"
1325                                name value (decode-char '=ucs value)
1326                                line-breaking))
1327                )
1328               ((eq name 'jisx0208-1978/4X)
1329                (insert (format "(%-18s . #x%04X)%s"
1330                                name value
1331                                line-breaking))
1332                )
1333               ((and
1334                 (not readable)
1335                 (not (eq name '->subsumptive))
1336                 (not (eq name '->uppercase))
1337                 (not (eq name '->lowercase))
1338                 (not (eq name '->titlecase))
1339                 (not (eq name '->canonical))
1340                 (not (eq name '->Bopomofo))
1341                 (not (eq name '->mistakable))
1342                 (not (eq name '->ideographic-variants))
1343                 (null (get-char-attribute
1344                        char (intern (format "%s*sources" name))))
1345                 (not (string-match "\\*sources$" (symbol-name name)))
1346                 (null (get-char-attribute
1347                        char (intern (format "%s*note" name))))
1348                 (not (string-match "\\*note$" (symbol-name name)))
1349                 (or (eq name '<-identical)
1350                     (eq name '<-uppercase)
1351                     (eq name '<-lowercase)
1352                     (eq name '<-titlecase)
1353                     (eq name '<-canonical)
1354                     (eq name '<-ideographic-variants)
1355                     ;; (eq name '<-synonyms)
1356                     (string-match "^<-synonyms" (symbol-name name))
1357                     (eq name '<-mistakable)
1358                     (when (string-match "^->" (symbol-name name))
1359                       (cond
1360                        ((string-match "^->fullwidth" (symbol-name name))
1361                         (not (and (consp value)
1362                                   (characterp (car value))
1363                                   (encode-char
1364                                    (car value) '=ucs 'defined-only)))
1365                         )
1366                        (t)))
1367                     ))
1368                )
1369               ((or (eq name 'ideographic-structure)
1370                    (eq name 'ideographic-combination)
1371                    ;; (eq name 'ideographic-)
1372                    (eq name '=decomposition)
1373                    (char-feature-base-name= '=decomposition name)
1374                    (char-feature-base-name= '=>decomposition name)
1375                    ;; (string-match "^=>*decomposition\\(@[^*]+\\)?$"
1376                    ;;               (symbol-name name))
1377                    (string-match "^\\(->\\|<-\\)[^*]*$" (symbol-name name))
1378                    (string-match "^\\(->\\|<-\\)[^*]*\\*sources$"
1379                                  (symbol-name name))
1380                    )
1381                (char-db-insert-relation-feature char name value
1382                                                 line-breaking
1383                                                 ccss readable))
1384               ;; ((memq name '(ideograph=
1385               ;;               original-ideograph-of
1386               ;;               ancient-ideograph-of
1387               ;;               vulgar-ideograph-of
1388               ;;               wrong-ideograph-of
1389               ;;               ;; simplified-ideograph-of
1390               ;;               ideographic-variants
1391               ;;               ;; ideographic-different-form-of
1392               ;;               ))
1393               ;;  (insert (format "(%-18s%s " name line-breaking))
1394               ;;  (setq lbs (concat "\n" (make-string (current-column) ?\ ))
1395               ;;        separator nil)
1396               ;;  (while (consp value)
1397               ;;    (setq cell (car value))
1398               ;;    (if (and (consp cell)
1399               ;;             (consp (car cell)))
1400               ;;        (progn
1401               ;;          (if separator
1402               ;;              (insert lbs))
1403               ;;          (char-db-insert-alist cell readable)
1404               ;;          (setq separator lbs))
1405               ;;      (if separator
1406               ;;          (insert separator))
1407               ;;      (insert (prin1-to-string cell))
1408               ;;      (setq separator " "))
1409               ;;    (setq value (cdr value)))
1410               ;;  (insert ")")
1411               ;;  (insert line-breaking))
1412               ((consp value)
1413                (insert (format "(%-18s " name))
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                           (setq ret (condition-case nil
1421                                         (find-char cell)
1422                                       (error nil))))
1423                      (progn
1424                        (setq rest cell
1425                              al nil
1426                              cal nil)
1427                        (while rest
1428                          (setq key (car (car rest)))
1429                          (if (find-charset key)
1430                              (setq cal (cons key cal))
1431                            (setq al (cons key al)))
1432                          (setq rest (cdr rest)))
1433                        (if separator
1434                            (insert lbs))
1435                        (insert-char-attributes ret
1436                                                readable
1437                                                al ; cal
1438                                                nil 'for-sub-node)
1439                        (setq separator lbs))
1440                    (setq ret (prin1-to-string cell))
1441                    (if separator
1442                        (if (< (+ (current-column)
1443                                  (length ret)
1444                                  (length separator))
1445                               76)
1446                            (insert separator)
1447                          (insert lbs)))
1448                    (insert ret)
1449                    (setq separator " "))
1450                  (setq value (cdr value)))
1451                (insert ")")
1452                (insert line-breaking))
1453               (t
1454                (insert (format "(%-18s" name))
1455                (setq ret (prin1-to-string value))
1456                (unless (< (+ (current-column)
1457                              (length ret)
1458                              3)
1459                           76)
1460                  (insert line-breaking))
1461                (insert " . " ret ")" line-breaking)
1462                ;; (insert (format "(%-18s . %S)%s"
1463                ;;                 name value
1464                ;;                 line-breaking))
1465                )
1466               ))
1467       (setq attributes (cdr attributes)))
1468     (insert ")")))
1469
1470 (defun insert-char-data (char &optional readable
1471                               attributes)
1472   (save-restriction
1473     (narrow-to-region (point)(point))
1474     (insert "(define-char
1475   '")
1476     (insert-char-attributes char readable attributes)
1477     (insert ")\n")
1478     (goto-char (point-min))
1479     (while (re-search-forward "[ \t]+$" nil t)
1480       (replace-match ""))
1481     ;; from tabify.
1482     (goto-char (point-min))
1483     (while (re-search-forward "[ \t][ \t][ \t]*" nil t)
1484       (let ((column (current-column))
1485             (indent-tabs-mode t))
1486         (delete-region (match-beginning 0) (point))
1487         (indent-to column)))
1488     (goto-char (point-max))
1489     ;; (tabify (point-min)(point-max))
1490     ))
1491
1492 (defun insert-char-data-with-variant (char &optional printable
1493                                            no-ucs-unified
1494                                            script excluded-script)
1495   (insert-char-data char printable)
1496   (let ((variants (char-variants char))
1497         rest
1498         variant vs ret)
1499     (setq variants (sort variants #'<))
1500     (setq rest variants)
1501     (setq variants (cons char variants))
1502     (while rest
1503       (setq variant (car rest))
1504       (unless (get-char-attribute variant '<-subsumptive)
1505         (if (and (or (null script)
1506                      (null (setq vs (get-char-attribute variant 'script)))
1507                      (memq script vs))
1508                  (or (null excluded-script)
1509                      (null (setq vs (get-char-attribute variant 'script)))
1510                      (not (memq excluded-script vs))))
1511             (unless (and no-ucs-unified (get-char-attribute variant '=ucs))
1512               (insert-char-data variant printable)
1513               (if (setq ret (char-variants variant))
1514                   (while ret
1515                     (or (memq (car ret) variants)
1516                         ;; (get-char-attribute (car ret) '<-subsumptive)
1517                         (setq rest (nconc rest (list (car ret)))))
1518                     (setq ret (cdr ret)))))))
1519       (setq rest (cdr rest)))))
1520
1521 (defun insert-char-range-data (min max &optional script excluded-script)
1522   (let ((code min)
1523         char)
1524     (while (<= code max)
1525       (setq char (decode-char '=ucs code))
1526       (if (encode-char char '=ucs 'defined-only)
1527           (insert-char-data-with-variant char nil 'no-ucs-unified
1528                                          script excluded-script))
1529       (setq code (1+ code)))))
1530
1531 (defun write-char-range-data-to-file (min max file
1532                                           &optional script excluded-script)
1533   (let ((coding-system-for-write char-db-file-coding-system))
1534     (with-temp-buffer
1535       (insert (format ";; -*- coding: %s -*-\n"
1536                       char-db-file-coding-system))
1537       (insert-char-range-data min max script excluded-script)
1538       (write-region (point-min)(point-max) file))))
1539
1540 (defvar what-character-original-window-configuration)
1541
1542 ;;;###autoload
1543 (defun what-char-definition (char)
1544   (interactive (list (char-after)))
1545   (let ((buf (get-buffer-create "*Character Description*"))
1546         (the-buf (current-buffer))
1547         (win-conf (current-window-configuration)))
1548     (pop-to-buffer buf)
1549     (make-local-variable 'what-character-original-window-configuration)
1550     (setq what-character-original-window-configuration win-conf)
1551     (setq buffer-read-only nil)
1552     (erase-buffer)
1553     (condition-case err
1554         (progn
1555           (insert-char-data-with-variant char 'printable)
1556           (unless (char-attribute-alist char)
1557             (insert (format ";; = %c\n"
1558                             (let* ((rest (split-char char))
1559                                    (ccs (pop rest))
1560                                    (code (pop rest)))
1561                               (while rest
1562                                 (setq code (logior (lsh code 8)
1563                                                    (pop rest))))
1564                               (decode-char ccs code)))))
1565           ;; (char-db-update-comment)
1566           (set-buffer-modified-p nil)
1567           (view-mode the-buf (lambda (buf)
1568                                (set-window-configuration
1569                                 what-character-original-window-configuration)
1570                                ))
1571           (goto-char (point-min)))
1572       (error (progn
1573                (set-window-configuration
1574                 what-character-original-window-configuration)
1575                (signal (car err) (cdr err)))))))
1576
1577
1578 ;;; @ end
1579 ;;;
1580
1581 (provide 'char-db-util)
1582
1583 ;;; char-db-util.el ends here