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 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     ;; ==mj
 172     ;; ==adobe-japan1-0
 173     ;; ==adobe-japan1-1
 174     ;; ==adobe-japan1-2
 175     ;; ==adobe-japan1-3
 176     ;; ==adobe-japan1-4
 177     ;; ==adobe-japan1-5
 178     ;; ==adobe-japan1-6
 179     ==jis-x0208
 180     ==jis-x0213-1
 181     ==jis-x0213-2
 182     ==hanyo-denshi/ja
 183     ==hanyo-denshi/jb
 184     ==hanyo-denshi/jc
 185     ==hanyo-denshi/ft
 186     ==hanyo-denshi/ib
 187     ==hanyo-denshi/hg
 188     ==hanyo-denshi/jt
 189     ==hanyo-denshi/ks
 190     =gt
 191     =gt-k
 192     =daikanwa
 193     =daikanwa@rev2
 194     =daikanwa@rev1
 195     =daikanwa/+p
 196     ==daikanwa
 197     =cbeta
 198     ideograph-hanziku-1
 199     ideograph-hanziku-2
 200     ideograph-hanziku-3
 201     ideograph-hanziku-4
 202     ideograph-hanziku-5
 203     ideograph-hanziku-6
 204     ideograph-hanziku-7
 205     ideograph-hanziku-8
 206     ideograph-hanziku-9
 207     ideograph-hanziku-10
 208     ideograph-hanziku-11
 209     ideograph-hanziku-12
 210     ;; =>>>adobe-japan1-0
 211     ;; =>>>adobe-japan1-1
 212     ;; =>>>adobe-japan1-2
 213     ;; =>>>adobe-japan1-3
 214     ;; =>>>adobe-japan1-4
 215     ;; =>>>adobe-japan1-5
 216     ;; =>>>adobe-japan1-6
 217     ;; =>>>jis-x0208
 218     ;; =>>>jis-x0213-1
 219     ;; =>>>jis-x0213-2
 220     ;; =>>>hanyo-denshi/ja
 221     ;; =>>>hanyo-denshi/jb
 222     ;; =>>>hanyo-denshi/jc
 223     ;; =>>>hanyo-denshi/ft
 224     ;; =>>>hanyo-denshi/ib
 225     ;; =>>>hanyo-denshi/hg
 226     ;; =>>>hanyo-denshi/jt
 227     ;; =>>>hanyo-denshi/ks
 228     ;; =>>>gt
 229     =>>adobe-japan1-0
 230     =>>adobe-japan1-1
 231     =>>adobe-japan1-2
 232     =>>adobe-japan1-3
 233     =>>adobe-japan1-4
 234     =>>adobe-japan1-5
 235     =>>adobe-japan1-6
 236     =>>jis-x0208
 237     =>>jis-x0213-1
 238     =>>jis-x0213-1@2000
 239     =>>jis-x0213-1@2004
 240     =>>jis-x0213-2
 241     =>>jis-x0208@1978
 242     =>>hanyo-denshi/ft
 243     =>>hanyo-denshi/jt
 244     =>>hanyo-denshi/ks
 245     =>>gt
 246     =>>daikanwa
 247     =>>cbeta
 248     =+>jis-x0208
 249     =+>jis-x0213-1
 250     =+>jis-x0213-2
 251     =+>adobe-japan1-0
 252     =+>adobe-japan1-1
 253     =+>adobe-japan1-2
 254     =+>adobe-japan1-3
 255     =+>adobe-japan1-4
 256     =+>adobe-japan1-5
 257     =+>adobe-japan1-6
 258     =+>jis-x0208@1978
 259     =>jis-x0208
 260     =>jis-x0208@1997
 261     =>jis-x0213-1
 262     =>jis-x0213-1@2000
 263     =>jis-x0213-1@2004
 264     =>jis-x0213-2
 265     ==>ucs@bucs
 266     =>iwds-1
 267     =>ucs@hanyo-denshi
 268     =>ucs@iso
 269     =>ucs@unicode
 270     =>ucs@jis
 271     =>ucs@cns
 272     =>ucs@ks
 273     =+>ucs@iso
 274     =+>ucs@unicode
 275     =+>ucs@jis
 276     =+>ucs@jis/1990
 277     =+>ucs@cns
 278     =+>ucs@ks
 279     =>>ucs@iso
 280     =>>ucs@unicode
 281     =>>ucs@jis
 282     =>>ucs@cns
 283     =>>>ucs@iso
 284     =>>>ucs@unicode
 285     ==ucs@iso
 286     ==ucs@unicode
 287     ;; ==ucs@cns
 288     ==gb2312
 289     ==ks-x1001
 290     ==cns11643-1
 291     ==cns11643-2
 292     ==cns11643-3
 293     ==cns11643-4
 294     ==cns11643-5
 295     ==cns11643-6
 296     ==cns11643-7
 297     ==gt
 298     ==jis-x0208@1990
 299     ;; ==jis-x0208@1983
 300     ==jis-x0208@1978
 301     ==gt-k
 302     =ucs@iso
 303     =ucs@unicode
 304     =ucs@cns
 305     ==big5-cdp
 306     ==cbeta
 307     =>>big5-cdp
 308     =>>gt-k
 309     =+>gt
 310     =>gt
 311     =>big5-cdp
 312     =>daikanwa
 313     =>daikanwa/ho
 314     =>cns11643-7
 315     =big5
 316     =big5-eten
 317     =>gt-k
 318     =zinbun-oracle
 319     =>zinbun-oracle
 320     =ruimoku-v6
 321     =>>ruimoku-v6
 322     ==ruimoku-v6
 323     =jef-china3
 324     =>cbeta
 325     =shinjigen
 326     =big5-cdp-var-3
 327     =big5-cdp-var-5
 328     =>ucs@iwds-1))
 329
 330
 331 ;;; @ char-db formatters
 332 ;;;
 333
 334 (defun char-db-make-char-spec (char)
 335   (let (ret char-spec)
 336     (cond ((characterp char)
 337            (cond ((and (setq ret (encode-char char '=ucs 'defined-only))
 338                        (not (and (<= #xE000 ret)(<= ret #xF8FF))))
 339                   (setq char-spec (list (cons '=ucs ret)))
 340                   (cond ((setq ret (get-char-attribute char 'name))
 341                          (setq char-spec (cons (cons 'name ret) char-spec))
 342                          )
 343                         ((setq ret (get-char-attribute char 'name*))
 344                          (setq char-spec (cons (cons 'name* ret) char-spec))
 345                          ))
 346                   )
 347                  ((encode-char char '=mj 'defined-only)
 348                   (setq char-spec nil)
 349                   (dolist (ccs (charset-list))
 350                     (if (and (or (eq ccs '=mj)
 351                                  ;; (eq (charset-property ccs 'iso-ir) 177)
 352                                  (string-match "=ucs@" (symbol-name ccs))
 353                                  )
 354                              (setq ccs (charset-name ccs))
 355                              (null (assq ccs char-spec))
 356                              (setq ret (encode-char char ccs 'defined-only)))
 357                         (setq char-spec (cons (cons ccs ret) char-spec))))
 358                   )
 359                  ((encode-char char '==mj 'defined-only)
 360                   (setq char-spec nil)
 361                   (dolist (ccs (charset-list))
 362                     (if (and (or (eq ccs '==mj)
 363                                  ;; (eq (charset-property ccs 'iso-ir) 177)
 364                                  (string-match "=ucs@" (symbol-name ccs))
 365                                  )
 366                              (setq ccs (charset-name ccs))
 367                              (null (assq ccs char-spec))
 368                              (setq ret (encode-char char ccs 'defined-only)))
 369                         (setq char-spec (cons (cons ccs ret) char-spec))))
 370                   )
 371                  ((encode-char char '=adobe-japan1 'defined-only)
 372                   (setq char-spec nil)
 373                   (dolist (ccs (charset-list))
 374                     (if (and (or (memq ccs
 375                                        '(=adobe-japan1-0
 376                                          =adobe-japan1-1
 377                                          =adobe-japan1-2
 378                                          =adobe-japan1-3
 379                                          =adobe-japan1-4
 380                                          =adobe-japan1-5
 381                                          =adobe-japan1-6
 382                                          ))
 383                                  ;; (eq (charset-property ccs 'iso-ir) 177)
 384                                  (string-match "=ucs@" (symbol-name ccs))
 385                                  )
 386                              (setq ccs (charset-name ccs))
 387                              (null (assq ccs char-spec))
 388                              (setq ret (encode-char char ccs 'defined-only)))
 389                         (setq char-spec (cons (cons ccs ret) char-spec))))
 390                   )
 391                  ((encode-char char '==adobe-japan1 'defined-only)
 392                   (setq char-spec nil)
 393                   (dolist (ccs (charset-list))
 394                     (if (and (or (memq ccs
 395                                        '(==adobe-japan1-0
 396                                          ==adobe-japan1-1
 397                                          ==adobe-japan1-2
 398                                          ==adobe-japan1-3
 399                                          ==adobe-japan1-4
 400                                          ==adobe-japan1-5
 401                                          ==adobe-japan1-6
 402                                          ))
 403                                  ;; (eq (charset-property ccs 'iso-ir) 177)
 404                                  (string-match "=ucs@" (symbol-name ccs))
 405                                  )
 406                              (setq ccs (charset-name ccs))
 407                              (null (assq ccs char-spec))
 408                              (setq ret (encode-char char ccs 'defined-only)))
 409                         (setq char-spec (cons (cons ccs ret) char-spec))))
 410                   )
 411                  ((setq ret
 412                         (catch 'tag
 413                           (let ((rest char-db-coded-charset-priority-list)
 414                                 ccs)
 415                             (while rest
 416                               (setq ccs (charset-name
 417                                          (find-charset (car rest))))
 418                               (if (setq ret
 419                                         (encode-char char ccs
 420                                                      'defined-only))
 421                                   (throw 'tag (cons ccs ret)))
 422                               (setq rest (cdr rest))))))
 423                   (setq char-spec (list ret))
 424                   (dolist (ccs (delq (car ret) (charset-list)))
 425                     (if (and (or (charset-iso-final-char ccs)
 426                                  (memq ccs
 427                                        '(=daikanwa
 428                                          =daikanwa@rev2
 429                                          ;; =gt-k
 430                                          =jis-x0208@1997
 431                                          ))
 432                                  (eq (charset-property ccs 'iso-ir) 177)
 433                                  ;; (string-match "=ucs@" (symbol-name ccs))
 434                                  )
 435                              (setq ccs (charset-name ccs))
 436                              (null (assq ccs char-spec))
 437                              (setq ret (encode-char char ccs 'defined-only)))
 438                         (setq char-spec (cons (cons ccs ret) char-spec))))
 439                   (if (null char-spec)
 440                       (setq char-spec (split-char char)))
 441                   (cond ((setq ret (get-char-attribute char 'name))
 442                          (setq char-spec (cons (cons 'name ret) char-spec))
 443                          )
 444                         ((setq ret (get-char-attribute char 'name*))
 445                          (setq char-spec (cons (cons 'name* ret) char-spec))
 446                          ))
 447                   )
 448                  ((setq ret (get-char-attribute
 449                              char 'ideographic-combination))
 450                   (setq char-spec
 451                         (cons (cons 'ideographic-combination ret)
 452                               char-spec))
 453                   ))
 454            char-spec)
 455           ((consp char)
 456            char))))
 457
 458 (defun char-db-insert-char-spec (char &optional readable column
 459                                       required-features)
 460   (unless column
 461     (setq column (current-column)))
 462   (let (char-spec temp-char)
 463     (setq char-spec (char-db-make-char-spec char))
 464     (unless (or (characterp char) ; char
 465                 (condition-case nil
 466                     (setq char (find-char char-spec))
 467                   (error nil)))
 468       ;; define temporary character
 469       ;;   Current implementation is dirty.
 470       (setq temp-char (define-char (cons '(ideograph-daikanwa . 0)
 471                                          char-spec)))
 472       (remove-char-attribute temp-char 'ideograph-daikanwa)
 473       (setq char temp-char))
 474     (insert-char-attributes char
 475                             readable
 476                             (union (mapcar #'car char-spec)
 477                                    required-features)
 478                             nil 'for-sub-node)
 479     (when temp-char
 480       ;; undefine temporary character
 481       ;;   Current implementation is dirty.
 482       (setq char-spec (char-attribute-alist temp-char))
 483       (while char-spec
 484         (remove-char-attribute temp-char (car (car char-spec)))
 485         (setq char-spec (cdr char-spec))))))
 486
 487 (defun char-db-insert-alist (alist &optional readable column)
 488   (unless column
 489     (setq column (current-column)))
 490   (let ((line-breaking
 491          (concat "\n" (make-string (1+ column) ?\ )))
 492         name value
 493         ret al ; cal
 494         key
 495         lbs cell rest separator)
 496     (insert "(")
 497     (while alist
 498       (setq name (car (car alist))
 499             value (cdr (car alist)))
 500       (cond ((eq name 'char)
 501              (insert "(char . ")
 502              (if (setq ret (condition-case nil
 503                                (find-char value)
 504                              (error nil)))
 505                  (progn
 506                    (setq al nil
 507                          ;; cal nil
 508                          )
 509                    (while value
 510                      (setq key (car (car value)))
 511                      ;; (if (find-charset key)
 512                      ;;     (setq cal (cons key cal))
 513                      (setq al (cons key al))
 514                      ;; )
 515                      (setq value (cdr value)))
 516                    (insert-char-attributes ret
 517                                            readable
 518                                            (or al 'none) ; cal
 519                                            nil 'for-sub-node))
 520                (insert (prin1-to-string value)))
 521              (insert ")")
 522              (insert line-breaking))
 523             ((consp value)
 524              (insert (format "(%-18s " name))
 525              (setq lbs (concat "\n" (make-string (current-column) ?\ )))
 526              (while (consp value)
 527                (setq cell (car value))
 528                (if (and (consp cell)
 529                         (consp (car cell))
 530                         (setq ret (condition-case nil
 531                                       (find-char cell)
 532                                     (error nil)))
 533                         )
 534                    (progn
 535                      (setq rest cell
 536                            al nil
 537                            ;; cal nil
 538                            )
 539                      (while rest
 540                        (setq key (car (car rest)))
 541                        ;; (if (find-charset key)
 542                        ;;     (setq cal (cons key cal))
 543                        (setq al (cons key al))
 544                        ;; )
 545                        (setq rest (cdr rest)))
 546                      (if separator
 547                          (insert lbs))
 548                      (insert-char-attributes ret
 549                                              readable
 550                                              al ; cal
 551                                              nil 'for-sub-node)
 552                      (setq separator lbs))
 553                  (if separator
 554                      (insert separator))
 555                  (insert (prin1-to-string cell))
 556                  (setq separator " "))
 557                (setq value (cdr value)))
 558              (insert ")")
 559              (insert line-breaking))
 560             (t
 561              (insert (format "(%-18s . %S)%s"
 562                              name value
 563                              line-breaking))))
 564       (setq alist (cdr alist))))
 565   (insert ")"))
 566
 567 (defun char-db-insert-char-reference (plist &optional readable column)
 568   (unless column
 569     (setq column (current-column)))
 570   (let ((line-breaking
 571          (concat "\n" (make-string (1+ column) ?\ )))
 572         (separator "")
 573         name value)
 574     (insert "(")
 575     (while plist
 576       (setq name (pop plist))
 577       (setq value (pop plist))
 578       (cond ((eq name :char)
 579              (insert separator)
 580              (insert ":char\t")
 581              (cond ((numberp value)
 582                     (setq value (decode-char '=ucs value)))
 583                    ;; ((consp value)
 584                    ;;  (setq value (or (find-char value)
 585                    ;;                  value)))
 586                    )
 587              (char-db-insert-char-spec value readable)
 588              (insert line-breaking)
 589              (setq separator ""))
 590             ((eq name :radical)
 591              (insert (format "%s%s\t%d ; %c%s"
 592                              separator
 593                              name value
 594                              (ideographic-radical value)
 595                              line-breaking))
 596              (setq separator ""))
 597             (t
 598              (insert (format "%s%s\t%S" separator name value))
 599              (setq separator line-breaking)))
 600       ))
 601   (insert ")"))
 602
 603 (defun char-db-decode-isolated-char (ccs code-point)
 604   (let (ret)
 605     (setq ret
 606           (cond ((eq ccs 'arabic-iso8859-6)
 607                  (decode-char ccs code-point))
 608                 ;; ((eq ccs '=gt)
 609                 ;;  (decode-builtin-char '==gt code-point))
 610                 ((and (memq ccs '(=gt-pj-1
 611                                   =gt-pj-2
 612                                   =gt-pj-3
 613                                   =gt-pj-4
 614                                   =gt-pj-5
 615                                   =gt-pj-6
 616                                   =gt-pj-7
 617                                   =gt-pj-8
 618                                   =gt-pj-9
 619                                   =gt-pj-10
 620                                   =gt-pj-11))
 621                       (setq ret (decode-char ccs code-point))
 622                       (setq ret (encode-char ret '=gt 'defined-only)))
 623                  (decode-builtin-char '=gt ret))
 624                 (t
 625                  (decode-builtin-char ccs code-point))))
 626     (cond ((null ret)
 627            (or (decode-char ccs code-point)
 628                (define-char (list (cons ccs code-point)))))
 629           ((and (<= 0 (char-int ret))
 630                 (<= (char-int ret) #x1F))
 631            (decode-char '=ucs (+ #x2400 (char-int ret))))
 632           ((= (char-int ret) #x7F)
 633            ?\u2421)
 634           (t ret))))
 635
 636 (defvar char-db-convert-obsolete-format t)
 637
 638 (defun char-db-insert-ccs-feature (name value line-breaking)
 639   (cond
 640    ((integerp value)
 641     (insert
 642      (format
 643       (cond
 644        ((memq name '(=>iwds-1
 645                      ==shinjigen
 646                      =shinjigen
 647                      =shinjigen@1ed ==shinjigen@1ed
 648                      =shinjigen@rev ==shinjigen@rev
 649                      =shinjigen/+p@rev ==shinjigen/+p@rev
 650                      ===daikanwa/ho ==daikanwa/ho
 651                      =daikanwa/ho =>>daikanwa/ho =>daikanwa/ho))
 652         "(%-18s .  %04d)\t; %c")
 653        ((eq name '=shinjigen@1ed/24pr)
 654         "(%-18s . %04d)\t; %c")
 655        ((or (memq name '(===daikanwa
 656                          ==daikanwa =daikanwa =>>daikanwa =>daikanwa
 657                          =daikanwa@rev1 =daikanwa@rev2
 658                          =daikanwa/+p ==daikanwa/+p ===daikanwa/+p
 659                          =>>daikanwa/+p
 660                          =daikanwa/+2p =>>daikanwa/+2p
 661                          =gt ==gt ===gt
 662                          =>>gt =+>gt =>gt
 663                          =gt-k ==gt-k ===gt-k
 664                          =>>gt-k =>gt-k
 665                          =adobe-japan1-0 ==adobe-japan1-0 ===adobe-japan1-0
 666                          =adobe-japan1-1 ==adobe-japan1-1 ===adobe-japan1-1
 667                          =adobe-japan1-2 ==adobe-japan1-2 ===adobe-japan1-2
 668                          =adobe-japan1-3 ==adobe-japan1-3 ===adobe-japan1-3
 669                          =adobe-japan1-4 ==adobe-japan1-4 ===adobe-japan1-4
 670                          =adobe-japan1-5 ==adobe-japan1-5 ===adobe-japan1-5
 671                          =adobe-japan1-6 ==adobe-japan1-6 ===adobe-japan1-6
 672                          =>>adobe-japan1-0 =+>adobe-japan1-0
 673                          =>>adobe-japan1-1 =+>adobe-japan1-1
 674                          =>>adobe-japan1-2 =+>adobe-japan1-2
 675                          =>>adobe-japan1-3 =+>adobe-japan1-3
 676                          =>>adobe-japan1-4 =+>adobe-japan1-4
 677                          =>>adobe-japan1-5 =+>adobe-japan1-5
 678                          =>>adobe-japan1-6 =+>adobe-japan1-6
 679                          =>cbeta =cbeta =>>cbeta ==cbeta ===cbeta
 680                          =zinbun-oracle =>zinbun-oracle
 681                          ===hng-khi ===hng-khm ===hng-kak
 682                          ===hng-sys ===hng-tsu
 683                          ===hng-hod ===hng-gok ===hng-k24
 684                          ===hng-kcj ===hng-kbk ===hng-sik
 685                          ===hng-skk ===hng-kyu ===hng-ksk ===hng-wan
 686                          ===hng-kss))
 687             ;; (string-match "^=adobe-" (symbol-name name))
 688             )
 689         "(%-18s . %05d)\t; %c")
 690        ((memq name '(=hanyo-denshi/ks
 691                      ==hanyo-denshi/ks ===hanyo-denshi/ks
 692                      =>>hanyo-denshi/ks
 693                      =koseki
 694                      =mj ==mj ===mj =>>mj
 695                      =zihai mojikyo))
 696         "(%-18s . %06d)\t; %c")
 697        ((>= (charset-dimension name) 2)
 698         "(%-18s . #x%04X)\t; %c")
 699        (t
 700         "(%-18s . #x%02X)\t; %c"))
 701       name
 702       (if (= (charset-iso-graphic-plane name) 1)
 703           (logior value
 704                   (cond ((= (charset-dimension name) 1)
 705                          #x80)
 706                         ((= (charset-dimension name) 2)
 707                          #x8080)
 708                         ((= (charset-dimension name) 3)
 709                          #x808080)
 710                         (t 0)))
 711         value)
 712       (char-db-decode-isolated-char name value)))
 713     (if (and (= (charset-chars name) 94)
 714              (= (charset-dimension name) 2))
 715         (insert (format " [%02d-%02d]"
 716                         (- (lsh value -8) 32)
 717                         (- (logand value 255) 32))))
 718     )
 719    (t
 720     (insert (format "(%-18s . %s)" name value))
 721     ))
 722   (insert line-breaking))
 723
 724 (defun char-db-insert-relation-feature (char name value line-breaking
 725                                              ccss readable)
 726   (insert (format "(%-18s%s " name line-breaking))
 727   (let ((lbs (concat "\n" (make-string (current-column) ?\ )))
 728         separator cell sources required-features
 729         ret)
 730     (while (consp value)
 731       (setq cell (car value))
 732       (if (integerp cell)
 733           (setq cell (decode-char '=ucs cell)))
 734       (cond
 735        ((eq name '->subsumptive)
 736         (when (or (not (some (lambda (atr)
 737                                (get-char-attribute cell atr))
 738                              char-db-ignored-attributes))
 739                   (some (lambda (ccs)
 740                           (encode-char cell ccs 'defined-only))
 741                         ccss))
 742           (if separator
 743               (insert lbs))
 744           (let ((char-db-ignored-attributes
 745                  (cons '<-subsumptive
 746                        char-db-ignored-attributes)))
 747             (insert-char-attributes cell readable nil nil 'for-sub-node))
 748           (setq separator lbs))
 749         )
 750        ((characterp cell)
 751         (setq sources
 752               (get-char-attribute
 753                char (intern (format "%s*sources" name))))
 754         (setq required-features nil)
 755         (dolist (source sources)
 756           (cond
 757            ((memq source '(JP
 758                            JP/Jouyou
 759                            shinjigen shinjigen@1ed shinjigen@rev))
 760             (setq required-features
 761                   (union required-features
 762                          '(=jis-x0208
 763                            =jis-x0208@1990
 764                            =jis-x0213-1@2000
 765                            =jis-x0213-1@2004
 766                            =jis-x0213-2
 767                            =jis-x0212
 768                            =jis-x0208@1983
 769                            =jis-x0208@1978
 770                            =shinjigen
 771                            =shinjigen@1ed
 772                            =shinjigen@rev
 773                            =shinjigen/+p@rev))))
 774            ((eq source 'CN)
 775             (setq required-features
 776                   (union required-features
 777                          '(=gb2312
 778                            =gb12345
 779                            =iso-ir165)))))
 780           (cond
 781            ((find-charset (setq ret (intern (format "=%s" source))))
 782             (setq required-features
 783                   (cons ret required-features)))
 784            (t (setq required-features
 785                     (cons source required-features)))))
 786         (cond ((string-match "@JP" (symbol-name name))
 787                (setq required-features
 788                      (union required-features
 789                             '(=jis-x0208
 790                               =jis-x0208@1990
 791                               =jis-x0213-1-2000
 792                               =jis-x0213-2-2000
 793                               =jis-x0212
 794                               =jis-x0208@1983
 795                               =jis-x0208@1978))))
 796               ((string-match "@CN" (symbol-name name))
 797                (setq required-features
 798                      (union required-features
 799                             '(=gb2312
 800                               =gb12345
 801                               =iso-ir165)))))
 802         (if separator
 803             (insert lbs))
 804         (if readable
 805             (insert (format "%S" cell))
 806           (char-db-insert-char-spec cell readable
 807                                     nil
 808                                     required-features))
 809         (setq separator lbs))
 810        ((consp cell)
 811         (if separator
 812             (insert lbs))
 813         (if (consp (car cell))
 814             (char-db-insert-char-spec cell readable)
 815           (char-db-insert-char-reference cell readable))
 816         (setq separator lbs))
 817        (t
 818         (if separator
 819             (insert separator))
 820         (insert (prin1-to-string cell))
 821         (setq separator " ")))
 822       (setq value (cdr value)))
 823     (insert ")")
 824     (insert line-breaking)))
 825
 826 (defun insert-char-attributes (char &optional readable attributes column
 827                                     for-sub-node)
 828   (unless column
 829     (setq column (current-column)))
 830   (let (name value ; has-long-ccs-name
 831         rest
 832         radical strokes
 833         (line-breaking
 834          (concat "\n" (make-string (1+ column) ?\ )))
 835         lbs cell separator ret
 836         key al cal
 837         dest-ccss ; sources required-features
 838         ccss)
 839     (let (atr-d)
 840       (setq attributes
 841             (sort (if attributes
 842                       (if (consp attributes)
 843                           (progn
 844                             (dolist (name attributes)
 845                               (unless (memq name char-db-ignored-attributes)
 846                                 (if (find-charset name)
 847                                     (push name ccss))
 848                                 (push name atr-d)))
 849                             atr-d))
 850                     (dolist (name (char-attribute-list))
 851                       (unless (memq name char-db-ignored-attributes)
 852                         (if (find-charset name)
 853                             (push name ccss))
 854                         (push name atr-d)))
 855                     atr-d)
 856                   #'char-attribute-name<)))
 857     (insert "(")
 858     (when (memq '<-subsumptive attributes)
 859       (when (or readable (not for-sub-node))
 860         (when (setq value (get-char-attribute char '<-subsumptive))
 861           (char-db-insert-relation-feature char '<-subsumptive value
 862                                            line-breaking
 863                                            ccss readable)))
 864       (setq attributes (delq '<-subsumptive attributes)))
 865     (when (and (memq '<-denotational attributes)
 866                (setq value (get-char-attribute char '<-denotational)))
 867       (char-db-insert-relation-feature char '<-denotational value
 868                                        line-breaking
 869                                        ccss readable)
 870       (setq attributes (delq '<-denotational attributes)))
 871     (when (and (memq '<-denotational@component attributes)
 872                (setq value (get-char-attribute char '<-denotational@component)))
 873       (char-db-insert-relation-feature char '<-denotational@component value
 874                                        line-breaking
 875                                        ccss readable)
 876       (setq attributes (delq '<-denotational@component attributes)))
 877     (when (and (memq 'name attributes)
 878                (setq value (get-char-attribute char 'name)))
 879       (insert (format
 880                (if (> (+ (current-column) (length value)) 48)
 881                    "(name . %S)%s"
 882                  "(name               . %S)%s")
 883                value line-breaking))
 884       (setq attributes (delq 'name attributes))
 885       )
 886     (when (and (memq 'name* attributes)
 887                (setq value (get-char-attribute char 'name*)))
 888       (insert (format
 889                (if (> (+ (current-column) (length value)) 48)
 890                    "(name* . %S)%s"
 891                  "(name*              . %S)%s")
 892                value line-breaking))
 893       (setq attributes (delq 'name* attributes))
 894       )
 895     (when (and (memq 'script attributes)
 896                (setq value (get-char-attribute char 'script)))
 897       (insert (format "(script\t\t%s)%s"
 898                       (mapconcat (function prin1-to-string)
 899                                  value " ")
 900                       line-breaking))
 901       (setq attributes (delq 'script attributes))
 902       )
 903     (dolist (name '(=>ucs =>ucs*))
 904       (when (and (memq name attributes)
 905                  (setq value (get-char-attribute char name)))
 906         (insert (format "(%-18s . #x%04X)\t; %c%s"
 907                         name value (decode-char '=ucs value)
 908                         line-breaking))
 909         (setq attributes (delq name attributes))))
 910     (dolist (name '(=>ucs@gb =>ucs@big5))
 911       (when (and (memq name attributes)
 912                  (setq value (get-char-attribute char name)))
 913         (insert (format "(%-18s . #x%04X)\t; %c%s"
 914                         name value
 915                         (decode-char (intern
 916                                       (concat "="
 917                                               (substring
 918                                                (symbol-name name) 2)))
 919                                      value)
 920                         line-breaking))
 921         (setq attributes (delq name attributes))
 922         ))
 923     ;; (dolist (name '(=>daikanwa))
 924     ;;   (when (and (memq name attributes)
 925     ;;              (setq value (get-char-attribute char name)))
 926     ;;     (insert
 927     ;;      (if (integerp value)
 928     ;;          (format "(%-18s . %05d)\t; %c%s"
 929     ;;                  name value (decode-char '=daikanwa value)
 930     ;;                  line-breaking)
 931     ;;        (format "(%-18s %s)\t; %c%s"
 932     ;;                name
 933     ;;                (mapconcat (function prin1-to-string)
 934     ;;                           value " ")
 935     ;;                (char-representative-of-daikanwa char)
 936     ;;                line-breaking)))
 937     ;;     (setq attributes (delq name attributes))))
 938     (when (and (memq 'general-category attributes)
 939                (setq value (get-char-attribute char 'general-category)))
 940       (insert (format
 941                "(general-category\t%s) ; %s%s"
 942                (mapconcat (lambda (cell)
 943                             (format "%S" cell))
 944                           value " ")
 945                (cond ((rassoc value unidata-normative-category-alist)
 946                       "Normative Category")
 947                      ((rassoc value unidata-informative-category-alist)
 948                       "Informative Category")
 949                      (t
 950                       "Unknown Category"))
 951                line-breaking))
 952       (setq attributes (delq 'general-category attributes))
 953       )
 954     (when (and (memq 'bidi-category attributes)
 955                (setq value (get-char-attribute char 'bidi-category)))
 956       (insert (format "(bidi-category\t. %S)%s"
 957                       value
 958                       line-breaking))
 959       (setq attributes (delq 'bidi-category attributes))
 960       )
 961     (unless (or (not (memq 'mirrored attributes))
 962                 (eq (setq value (get-char-attribute char 'mirrored 'empty))
 963                     'empty))
 964       (insert (format "(mirrored\t\t. %S)%s"
 965                       value
 966                       line-breaking))
 967       (setq attributes (delq 'mirrored attributes))
 968       )
 969     (cond
 970      ((and (memq 'decimal-digit-value attributes)
 971            (setq value (get-char-attribute char 'decimal-digit-value)))
 972       (insert (format "(decimal-digit-value . %S)%s"
 973                       value
 974                       line-breaking))
 975       (setq attributes (delq 'decimal-digit-value attributes))
 976       (when (and (memq 'digit-value attributes)
 977                  (setq value (get-char-attribute char 'digit-value)))
 978         (insert (format "(digit-value\t . %S)%s"
 979                         value
 980                         line-breaking))
 981         (setq attributes (delq 'digit-value attributes))
 982         )
 983       (when (and (memq 'numeric-value attributes)
 984                  (setq value (get-char-attribute char 'numeric-value)))
 985         (insert (format "(numeric-value\t . %S)%s"
 986                         value
 987                         line-breaking))
 988         (setq attributes (delq 'numeric-value attributes))
 989         )
 990       )
 991      (t
 992       (when (and (memq 'digit-value attributes)
 993                  (setq value (get-char-attribute char 'digit-value)))
 994         (insert (format "(digit-value\t. %S)%s"
 995                         value
 996                         line-breaking))
 997         (setq attributes (delq 'digit-value attributes))
 998         )
 999       (when (and (memq 'numeric-value attributes)
1000                  (setq value (get-char-attribute char 'numeric-value)))
1001         (insert (format "(numeric-value\t. %S)%s"
1002                         value
1003                         line-breaking))
1004         (setq attributes (delq 'numeric-value attributes))
1005         )))
1006     (when (and (memq 'iso-10646-comment attributes)
1007                (setq value (get-char-attribute char 'iso-10646-comment)))
1008       (insert (format "(iso-10646-comment\t. %S)%s"
1009                       value
1010                       line-breaking))
1011       (setq attributes (delq 'iso-10646-comment attributes))
1012       )
1013     (when (and (memq 'morohashi-daikanwa attributes)
1014                (setq value (get-char-attribute char 'morohashi-daikanwa)))
1015       (insert (format "(morohashi-daikanwa\t%s)%s"
1016                       (mapconcat (function prin1-to-string) value " ")
1017                       line-breaking))
1018       (setq attributes (delq 'morohashi-daikanwa attributes))
1019       )
1020     (setq radical nil
1021           strokes nil)
1022     (when (and (memq 'ideographic-radical attributes)
1023                (setq value (get-char-attribute char 'ideographic-radical)))
1024       (setq radical value)
1025       (insert (format "(ideographic-radical . %S)\t; %c%s"
1026                       radical
1027                       (ideographic-radical radical)
1028                       line-breaking))
1029       (setq attributes (delq 'ideographic-radical attributes))
1030       )
1031     (when (and (memq 'shuowen-radical attributes)
1032                (setq value (get-char-attribute char 'shuowen-radical)))
1033       (insert (format "(shuowen-radical\t. %S)\t; %c%s"
1034                       value
1035                       (shuowen-radical value)
1036                       line-breaking))
1037       (setq attributes (delq 'shuowen-radical attributes))
1038       )
1039     (let (key)
1040       (dolist (domain
1041                (append
1042                 char-db-feature-domains
1043                 (let (dest domain)
1044                   (dolist (feature (char-attribute-list))
1045                     (setq feature (symbol-name feature))
1046                     (when (string-match
1047                            "\\(radical\\|strokes\\)@\\([^@*]+\\)\\(\\*\\|$\\)"
1048                            feature)
1049                       (setq domain (intern (match-string 2 feature)))
1050                      (unless (memq domain dest)
1051                        (setq dest (cons domain dest)))))
1052                   (sort dest #'string<))))
1053         (setq key (intern (format "%s@%s" 'ideographic-radical domain)))
1054         (when (and (memq key attributes)
1055                    (setq value (get-char-attribute char key)))
1056           (setq radical value)
1057           (insert (format "(%s . %S)\t; %c%s"
1058                           key
1059                           radical
1060                           (ideographic-radical radical)
1061                           line-breaking))
1062           (setq attributes (delq key attributes))
1063           )
1064         (setq key (intern (format "%s@%s" 'ideographic-strokes domain)))
1065         (when (and (memq key attributes)
1066                    (setq value (get-char-attribute char key)))
1067           (setq strokes value)
1068           (insert (format "(%s . %S)%s"
1069                           key
1070                           strokes
1071                           line-breaking))
1072           (setq attributes (delq key attributes))
1073           )
1074         (setq key (intern (format "%s@%s" 'total-strokes domain)))
1075         (when (and (memq key attributes)
1076                    (setq value (get-char-attribute char key)))
1077           (insert (format "(%s       . %S)%s"
1078                           key
1079                           value
1080                           line-breaking))
1081           (setq attributes (delq key attributes))
1082           )
1083         (dolist (feature '(ideographic-radical
1084                            ideographic-strokes
1085                            total-strokes))
1086           (setq key (intern (format "%s@%s*sources" feature domain)))
1087           (when (and (memq key attributes)
1088                      (setq value (get-char-attribute char key)))
1089             (insert (format "(%s%s" key line-breaking))
1090             (dolist (cell value)
1091               (insert (format " %s" cell)))
1092             (insert ")")
1093             (insert line-breaking)
1094             (setq attributes (delq key attributes))
1095             ))
1096         ))
1097     (when (and (memq 'ideographic-strokes attributes)
1098                (setq value (get-char-attribute char 'ideographic-strokes)))
1099       (setq strokes value)
1100       (insert (format "(ideographic-strokes . %S)%s"
1101                       strokes
1102                       line-breaking))
1103       (setq attributes (delq 'ideographic-strokes attributes))
1104       )
1105     (when (and (memq 'kangxi-radical attributes)
1106                (setq value (get-char-attribute char 'kangxi-radical)))
1107       (unless (eq value radical)
1108         (insert (format "(kangxi-radical\t . %S)\t; %c%s"
1109                         value
1110                         (ideographic-radical value)
1111                         line-breaking))
1112         (or radical
1113             (setq radical value)))
1114       (setq attributes (delq 'kangxi-radical attributes))
1115       )
1116     (when (and (memq 'kangxi-strokes attributes)
1117                (setq value (get-char-attribute char 'kangxi-strokes)))
1118       (unless (eq value strokes)
1119         (insert (format "(kangxi-strokes\t . %S)%s"
1120                         value
1121                         line-breaking))
1122         (or strokes
1123             (setq strokes value)))
1124       (setq attributes (delq 'kangxi-strokes attributes))
1125       )
1126     (when (and (memq 'japanese-radical attributes)
1127                (setq value (get-char-attribute char 'japanese-radical)))
1128       (unless (eq value radical)
1129         (insert (format "(japanese-radical\t . %S)\t; %c%s"
1130                         value
1131                         (ideographic-radical value)
1132                         line-breaking))
1133         (or radical
1134             (setq radical value)))
1135       (setq attributes (delq 'japanese-radical attributes))
1136       )
1137     (when (and (memq 'japanese-strokes attributes)
1138                (setq value (get-char-attribute char 'japanese-strokes)))
1139       (unless (eq value strokes)
1140         (insert (format "(japanese-strokes\t . %S)%s"
1141                         value
1142                         line-breaking))
1143         (or strokes
1144             (setq strokes value)))
1145       (setq attributes (delq 'japanese-strokes attributes))
1146       )
1147     (when (and (memq 'cns-radical attributes)
1148                (setq value (get-char-attribute char 'cns-radical)))
1149       (insert (format "(cns-radical\t . %S)\t; %c%s"
1150                       value
1151                       (ideographic-radical value)
1152                       line-breaking))
1153       (setq attributes (delq 'cns-radical attributes))
1154       )
1155     (when (and (memq 'cns-strokes attributes)
1156                (setq value (get-char-attribute char 'cns-strokes)))
1157       (unless (eq value strokes)
1158         (insert (format "(cns-strokes\t . %S)%s"
1159                         value
1160                         line-breaking))
1161         (or strokes
1162             (setq strokes value)))
1163       (setq attributes (delq 'cns-strokes attributes))
1164       )
1165     ;; (when (and (memq 'shinjigen-1-radical attributes)
1166     ;;            (setq value (get-char-attribute char 'shinjigen-1-radical)))
1167     ;;   (unless (eq value radical)
1168     ;;     (insert (format "(shinjigen-1-radical . %S)\t; %c%s"
1169     ;;                     value
1170     ;;                     (ideographic-radical value)
1171     ;;                     line-breaking))
1172     ;;     (or radical
1173     ;;         (setq radical value)))
1174     ;;   (setq attributes (delq 'shinjigen-1-radical attributes))
1175     ;;   )
1176     (when (and (memq 'ideographic- attributes)
1177                (setq value (get-char-attribute char 'ideographic-)))
1178       (insert "(ideographic-       ")
1179       (setq lbs (concat "\n" (make-string (current-column) ?\ ))
1180             separator nil)
1181       (while (consp value)
1182         (setq cell (car value))
1183         (if (integerp cell)
1184             (setq cell (decode-char '=ucs cell)))
1185         (cond ((characterp cell)
1186                (if separator
1187                    (insert lbs))
1188                (if readable
1189                    (insert (format "%S" cell))
1190                  (char-db-insert-char-spec cell readable))
1191                (setq separator lbs))
1192               ((consp cell)
1193                (if separator
1194                    (insert lbs))
1195                (if (consp (car cell))
1196                    (char-db-insert-char-spec cell readable)
1197                  (char-db-insert-char-reference cell readable))
1198                (setq separator lbs))
1199               (t
1200                (if separator
1201                    (insert separator))
1202                (insert (prin1-to-string cell))
1203                (setq separator " ")))
1204         (setq value (cdr value)))
1205       (insert ")")
1206       (insert line-breaking)
1207       (setq attributes (delq 'ideographic- attributes)))
1208     (when (and (memq 'total-strokes attributes)
1209                (setq value (get-char-attribute char 'total-strokes)))
1210       (insert (format "(total-strokes       . %S)%s"
1211                       value
1212                       line-breaking))
1213       (setq attributes (delq 'total-strokes attributes))
1214       )
1215     (when (and (memq '->ideograph attributes)
1216                (setq value (get-char-attribute char '->ideograph)))
1217       (insert (format "(->ideograph\t%s)%s"
1218                       (mapconcat (lambda (code)
1219                                    (cond ((symbolp code)
1220                                           (symbol-name code))
1221                                          ((integerp code)
1222                                           (format "#x%04X" code))
1223                                          (t
1224                                           (format "%s %S"
1225                                                   line-breaking code))))
1226                                  value " ")
1227                       line-breaking))
1228       (setq attributes (delq '->ideograph attributes))
1229       )
1230     ;; (when (and (memq '->decomposition attributes)
1231     ;;            (setq value (get-char-attribute char '->decomposition)))
1232     ;;   (insert (format "(->decomposition\t%s)%s"
1233     ;;                   (mapconcat (lambda (code)
1234     ;;                                (cond ((symbolp code)
1235     ;;                                       (symbol-name code))
1236     ;;                                      ((characterp code)
1237     ;;                                       (if readable
1238     ;;                                           (format "%S" code)
1239     ;;                                         (format "#x%04X"
1240     ;;                                                 (char-int code))
1241     ;;                                         ))
1242     ;;                                      ((integerp code)
1243     ;;                                       (format "#x%04X" code))
1244     ;;                                      (t
1245     ;;                                       (format "%s%S" line-breaking code))))
1246     ;;                              value " ")
1247     ;;                   line-breaking))
1248     ;;   (setq attributes (delq '->decomposition attributes))
1249     ;;   )
1250     (if (equal (get-char-attribute char '->titlecase)
1251                (get-char-attribute char '->uppercase))
1252         (setq attributes (delq '->titlecase attributes)))
1253     (when (and (memq '->mojikyo attributes)
1254                (setq value (get-char-attribute char '->mojikyo)))
1255       (insert (format "(->mojikyo\t\t. %06d)\t; %c%s"
1256                       value (decode-char 'mojikyo value)
1257                       line-breaking))
1258       (setq attributes (delq '->mojikyo attributes))
1259       )
1260     (when (and (memq 'hanyu-dazidian-vol attributes)
1261                (setq value (get-char-attribute char 'hanyu-dazidian-vol)))
1262       (insert (format "(hanyu-dazidian-vol  . %d)%s"
1263                       value line-breaking))
1264       (setq attributes (delq 'hanyu-dazidian-vol attributes))
1265       )
1266     (when (and (memq 'hanyu-dazidian-page attributes)
1267                (setq value (get-char-attribute char 'hanyu-dazidian-page)))
1268       (insert (format "(hanyu-dazidian-page . %d)%s"
1269                       value line-breaking))
1270       (setq attributes (delq 'hanyu-dazidian-page attributes))
1271       )
1272     (when (and (memq 'hanyu-dazidian-char attributes)
1273                (setq value (get-char-attribute char 'hanyu-dazidian-char)))
1274       (insert (format "(hanyu-dazidian-char . %d)%s"
1275                       value line-breaking))
1276       (setq attributes (delq 'hanyu-dazidian-char attributes))
1277       )
1278     (unless readable
1279       (dolist (ignored '(composition
1280                          ->denotational <-subsumptive ->ucs-unified
1281                          ->ideographic-component-forms))
1282         (setq attributes (delq ignored attributes))))
1283     (while attributes
1284       (setq name (car attributes))
1285       (unless (eq (setq value (get-char-attribute char name 'value-is-empty))
1286                   'value-is-empty)
1287         (cond ((setq ret (find-charset name))
1288                (setq name (charset-name ret))
1289                (when (not (memq name dest-ccss))
1290                  (setq dest-ccss (cons name dest-ccss))
1291                  (char-db-insert-ccs-feature name value line-breaking))
1292                )
1293               ((string-match "^=>ucs@" (symbol-name name))
1294                (insert (format "(%-18s . #x%04X)\t; %c%s"
1295                                name value (decode-char '=ucs value)
1296                                line-breaking))
1297                )
1298               ((eq name 'jisx0208-1978/4X)
1299                (insert (format "(%-18s . #x%04X)%s"
1300                                name value
1301                                line-breaking))
1302                )
1303               ((and
1304                 (not readable)
1305                 (not (eq name '->subsumptive))
1306                 (not (eq name '->uppercase))
1307                 (not (eq name '->lowercase))
1308                 (not (eq name '->titlecase))
1309                 (not (eq name '->canonical))
1310                 (not (eq name '->Bopomofo))
1311                 (not (eq name '->mistakable))
1312                 (not (eq name '->ideographic-variants))
1313                 (null (get-char-attribute
1314                        char (intern (format "%s*sources" name))))
1315                 (not (string-match "\\*sources$" (symbol-name name)))
1316                 (null (get-char-attribute
1317                        char (intern (format "%s*note" name))))
1318                 (not (string-match "\\*note$" (symbol-name name)))
1319                 (or (eq name '<-identical)
1320                     (eq name '<-uppercase)
1321                     (eq name '<-lowercase)
1322                     (eq name '<-titlecase)
1323                     (eq name '<-canonical)
1324                     (eq name '<-ideographic-variants)
1325                     ;; (eq name '<-synonyms)
1326                     (string-match "^<-synonyms" (symbol-name name))
1327                     (eq name '<-mistakable)
1328                     (when (string-match "^->" (symbol-name name))
1329                       (cond
1330                        ((string-match "^->fullwidth" (symbol-name name))
1331                         (not (and (consp value)
1332                                   (characterp (car value))
1333                                   (encode-char
1334                                    (car value) '=ucs 'defined-only)))
1335                         )
1336                        (t)))
1337                     ))
1338                )
1339               ((or (eq name 'ideographic-structure)
1340                    (eq name 'ideographic-combination)
1341                    (eq name 'ideographic-)
1342                    (eq name '=decomposition)
1343                    (char-feature-base-name= '=decomposition name)
1344                    (char-feature-base-name= '=>decomposition name)
1345                    ;; (string-match "^=>*decomposition\\(@[^*]+\\)?$"
1346                    ;;               (symbol-name name))
1347                    (string-match "^\\(->\\|<-\\)[^*]*$" (symbol-name name))
1348                    (string-match "^\\(->\\|<-\\)[^*]*\\*sources$"
1349                                  (symbol-name name))
1350                    )
1351                (char-db-insert-relation-feature char name value
1352                                                 line-breaking
1353                                                 ccss readable))
1354               ((memq name '(ideograph=
1355                             original-ideograph-of
1356                             ancient-ideograph-of
1357                             vulgar-ideograph-of
1358                             wrong-ideograph-of
1359                             ;; simplified-ideograph-of
1360                             ideographic-variants
1361                             ;; ideographic-different-form-of
1362                             ))
1363                (insert (format "(%-18s%s " name line-breaking))
1364                (setq lbs (concat "\n" (make-string (current-column) ?\ ))
1365                      separator nil)
1366                (while (consp value)
1367                  (setq cell (car value))
1368                  (if (and (consp cell)
1369                           (consp (car cell)))
1370                      (progn
1371                        (if separator
1372                            (insert lbs))
1373                        (char-db-insert-alist cell readable)
1374                        (setq separator lbs))
1375                    (if separator
1376                        (insert separator))
1377                    (insert (prin1-to-string cell))
1378                    (setq separator " "))
1379                  (setq value (cdr value)))
1380                (insert ")")
1381                (insert line-breaking))
1382               ((consp value)
1383                (insert (format "(%-18s " name))
1384                (setq lbs (concat "\n" (make-string (current-column) ?\ ))
1385                      separator nil)
1386                (while (consp value)
1387                  (setq cell (car value))
1388                  (if (and (consp cell)
1389                           (consp (car cell))
1390                           (setq ret (condition-case nil
1391                                         (find-char cell)
1392                                       (error nil))))
1393                      (progn
1394                        (setq rest cell
1395                              al nil
1396                              cal nil)
1397                        (while rest
1398                          (setq key (car (car rest)))
1399                          (if (find-charset key)
1400                              (setq cal (cons key cal))
1401                            (setq al (cons key al)))
1402                          (setq rest (cdr rest)))
1403                        (if separator
1404                            (insert lbs))
1405                        (insert-char-attributes ret
1406                                                readable
1407                                                al ; cal
1408                                                nil 'for-sub-node)
1409                        (setq separator lbs))
1410                    (setq ret (prin1-to-string cell))
1411                    (if separator
1412                        (if (< (+ (current-column)
1413                                  (length ret)
1414                                  (length separator))
1415                               76)
1416                            (insert separator)
1417                          (insert lbs)))
1418                    (insert ret)
1419                    (setq separator " "))
1420                  (setq value (cdr value)))
1421                (insert ")")
1422                (insert line-breaking))
1423               (t
1424                (insert (format "(%-18s" name))
1425                (setq ret (prin1-to-string value))
1426                (unless (< (+ (current-column)
1427                              (length ret)
1428                              3)
1429                           76)
1430                  (insert line-breaking))
1431                (insert " . " ret ")" line-breaking)
1432                ;; (insert (format "(%-18s . %S)%s"
1433                ;;                 name value
1434                ;;                 line-breaking))
1435                )
1436               ))
1437       (setq attributes (cdr attributes)))
1438     (insert ")")))
1439
1440 (defun insert-char-data (char &optional readable
1441                               attributes)
1442   (save-restriction
1443     (narrow-to-region (point)(point))
1444     (insert "(define-char
1445   '")
1446     (insert-char-attributes char readable attributes)
1447     (insert ")\n")
1448     (goto-char (point-min))
1449     (while (re-search-forward "[ \t]+$" nil t)
1450       (replace-match ""))
1451     ;; from tabify.
1452     (goto-char (point-min))
1453     (while (re-search-forward "[ \t][ \t][ \t]*" nil t)
1454       (let ((column (current-column))
1455             (indent-tabs-mode t))
1456         (delete-region (match-beginning 0) (point))
1457         (indent-to column)))
1458     (goto-char (point-max))
1459     ;; (tabify (point-min)(point-max))
1460     ))
1461
1462 (defun insert-char-data-with-variant (char &optional printable
1463                                            no-ucs-unified
1464                                            script excluded-script)
1465   (insert-char-data char printable)
1466   (let ((variants (char-variants char))
1467         rest
1468         variant vs ret)
1469     (setq variants (sort variants #'<))
1470     (setq rest variants)
1471     (setq variants (cons char variants))
1472     (while rest
1473       (setq variant (car rest))
1474       (unless (get-char-attribute variant '<-subsumptive)
1475         (if (and (or (null script)
1476                      (null (setq vs (get-char-attribute variant 'script)))
1477                      (memq script vs))
1478                  (or (null excluded-script)
1479                      (null (setq vs (get-char-attribute variant 'script)))
1480                      (not (memq excluded-script vs))))
1481             (unless (and no-ucs-unified (get-char-attribute variant '=ucs))
1482               (insert-char-data variant printable)
1483               (if (setq ret (char-variants variant))
1484                   (while ret
1485                     (or (memq (car ret) variants)
1486                         ;; (get-char-attribute (car ret) '<-subsumptive)
1487                         (setq rest (nconc rest (list (car ret)))))
1488                     (setq ret (cdr ret)))))))
1489       (setq rest (cdr rest)))))
1490
1491 (defun insert-char-range-data (min max &optional script excluded-script)
1492   (let ((code min)
1493         char)
1494     (while (<= code max)
1495       (setq char (decode-char '=ucs code))
1496       (if (encode-char char '=ucs 'defined-only)
1497           (insert-char-data-with-variant char nil 'no-ucs-unified
1498                                          script excluded-script))
1499       (setq code (1+ code)))))
1500
1501 (defun write-char-range-data-to-file (min max file
1502                                           &optional script excluded-script)
1503   (let ((coding-system-for-write char-db-file-coding-system))
1504     (with-temp-buffer
1505       (insert (format ";; -*- coding: %s -*-\n"
1506                       char-db-file-coding-system))
1507       (insert-char-range-data min max script excluded-script)
1508       (write-region (point-min)(point-max) file))))
1509
1510 (defvar what-character-original-window-configuration)
1511
1512 ;;;###autoload
1513 (defun what-char-definition (char)
1514   (interactive (list (char-after)))
1515   (let ((buf (get-buffer-create "*Character Description*"))
1516         (the-buf (current-buffer))
1517         (win-conf (current-window-configuration)))
1518     (pop-to-buffer buf)
1519     (make-local-variable 'what-character-original-window-configuration)
1520     (setq what-character-original-window-configuration win-conf)
1521     (setq buffer-read-only nil)
1522     (erase-buffer)
1523     (condition-case err
1524         (progn
1525           (insert-char-data-with-variant char 'printable)
1526           (unless (char-attribute-alist char)
1527             (insert (format ";; = %c\n"
1528                             (let* ((rest (split-char char))
1529                                    (ccs (pop rest))
1530                                    (code (pop rest)))
1531                               (while rest
1532                                 (setq code (logior (lsh code 8)
1533                                                    (pop rest))))
1534                               (decode-char ccs code)))))
1535           ;; (char-db-update-comment)
1536           (set-buffer-modified-p nil)
1537           (view-mode the-buf (lambda (buf)
1538                                (set-window-configuration
1539                                 what-character-original-window-configuration)
1540                                ))
1541           (goto-char (point-min)))
1542       (error (progn
1543                (set-window-configuration
1544                 what-character-original-window-configuration)
1545                (signal (car err) (cdr err)))))))
1546
1547
1548 ;;; @ end
1549 ;;;
1550
1551 (provide 'char-db-util)
1552
1553 ;;; char-db-util.el ends here