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-jou
 682                          ===hng-khi ===hng-khm ===hng-kak
 683                          ===hng-sys ===hng-tsu
 684                          ===hng-hod ===hng-gok ===hng-k24
 685                          ===hng-kcj ===hng-kbk ===hng-sik
 686                          ===hng-skk ===hng-kyu ===hng-ksk ===hng-wan
 687                          ===hng-kss ===hng-kyo
 688                          ===hng-smk))
 689             ;; (string-match "^=adobe-" (symbol-name name))
 690             )
 691         "(%-18s . %05d)\t; %c")
 692        ((memq name '(=hanyo-denshi/ks
 693                      ==hanyo-denshi/ks ===hanyo-denshi/ks
 694                      =>>hanyo-denshi/ks
 695                      =koseki
 696                      =mj ==mj ===mj =>>mj
 697                      =zihai mojikyo))
 698         "(%-18s . %06d)\t; %c")
 699        ((>= (charset-dimension name) 2)
 700         "(%-18s . #x%04X)\t; %c")
 701        (t
 702         "(%-18s . #x%02X)\t; %c"))
 703       name
 704       (if (= (charset-iso-graphic-plane name) 1)
 705           (logior value
 706                   (cond ((= (charset-dimension name) 1)
 707                          #x80)
 708                         ((= (charset-dimension name) 2)
 709                          #x8080)
 710                         ((= (charset-dimension name) 3)
 711                          #x808080)
 712                         (t 0)))
 713         value)
 714       (char-db-decode-isolated-char name value)))
 715     (if (and (= (charset-chars name) 94)
 716              (= (charset-dimension name) 2))
 717         (insert (format " [%02d-%02d]"
 718                         (- (lsh value -8) 32)
 719                         (- (logand value 255) 32))))
 720     )
 721    (t
 722     (insert (format "(%-18s . %s)" name value))
 723     ))
 724   (insert line-breaking))
 725
 726 (defun char-db-insert-relation-feature (char name value line-breaking
 727                                              ccss readable)
 728   (insert (format "(%-18s%s " name line-breaking))
 729   (let ((lbs (concat "\n" (make-string (current-column) ?\ )))
 730         separator cell sources required-features
 731         ret)
 732     (while (consp value)
 733       (setq cell (car value))
 734       (if (integerp cell)
 735           (setq cell (decode-char '=ucs cell)))
 736       (cond
 737        ((eq name '->subsumptive)
 738         (when (or (not (some (lambda (atr)
 739                                (get-char-attribute cell atr))
 740                              char-db-ignored-attributes))
 741                   (some (lambda (ccs)
 742                           (encode-char cell ccs 'defined-only))
 743                         ccss))
 744           (if separator
 745               (insert lbs))
 746           (let ((char-db-ignored-attributes
 747                  (cons '<-subsumptive
 748                        char-db-ignored-attributes)))
 749             (insert-char-attributes cell readable nil nil 'for-sub-node))
 750           (setq separator lbs))
 751         )
 752        ((characterp cell)
 753         (setq sources
 754               (get-char-attribute
 755                char (intern (format "%s*sources" name))))
 756         (setq required-features nil)
 757         (dolist (source sources)
 758           (cond
 759            ((memq source '(JP
 760                            JP/Jouyou
 761                            shinjigen shinjigen@1ed shinjigen@rev))
 762             (setq required-features
 763                   (union required-features
 764                          '(=jis-x0208
 765                            =jis-x0208@1990
 766                            =jis-x0213-1@2000
 767                            =jis-x0213-1@2004
 768                            =jis-x0213-2
 769                            =jis-x0212
 770                            =jis-x0208@1983
 771                            =jis-x0208@1978
 772                            =shinjigen
 773                            =shinjigen@1ed
 774                            =shinjigen@rev
 775                            =shinjigen/+p@rev))))
 776            ((eq source 'CN)
 777             (setq required-features
 778                   (union required-features
 779                          '(=gb2312
 780                            =gb12345
 781                            =iso-ir165)))))
 782           (cond
 783            ((find-charset (setq ret (intern (format "=%s" source))))
 784             (setq required-features
 785                   (cons ret required-features)))
 786            (t (setq required-features
 787                     (cons source required-features)))))
 788         (cond ((string-match "@JP" (symbol-name name))
 789                (setq required-features
 790                      (union required-features
 791                             '(=jis-x0208
 792                               =jis-x0208@1990
 793                               =jis-x0213-1-2000
 794                               =jis-x0213-2-2000
 795                               =jis-x0212
 796                               =jis-x0208@1983
 797                               =jis-x0208@1978))))
 798               ((string-match "@CN" (symbol-name name))
 799                (setq required-features
 800                      (union required-features
 801                             '(=gb2312
 802                               =gb12345
 803                               =iso-ir165)))))
 804         (if separator
 805             (insert lbs))
 806         (if readable
 807             (insert (format "%S" cell))
 808           (char-db-insert-char-spec cell readable
 809                                     nil
 810                                     required-features))
 811         (setq separator lbs))
 812        ((consp cell)
 813         (if separator
 814             (insert lbs))
 815         (if (consp (car cell))
 816             (char-db-insert-char-spec cell readable)
 817           (char-db-insert-char-reference cell readable))
 818         (setq separator lbs))
 819        (t
 820         (if separator
 821             (insert separator))
 822         (insert (prin1-to-string cell))
 823         (setq separator " ")))
 824       (setq value (cdr value)))
 825     (insert ")")
 826     (insert line-breaking)))
 827
 828 (defun insert-char-attributes (char &optional readable attributes column
 829                                     for-sub-node)
 830   (unless column
 831     (setq column (current-column)))
 832   (let (name value ; has-long-ccs-name
 833         rest
 834         radical strokes
 835         (line-breaking
 836          (concat "\n" (make-string (1+ column) ?\ )))
 837         lbs cell separator ret
 838         key al cal
 839         dest-ccss ; sources required-features
 840         ccss)
 841     (let (atr-d)
 842       (setq attributes
 843             (sort (if attributes
 844                       (if (consp attributes)
 845                           (progn
 846                             (dolist (name attributes)
 847                               (unless (memq name char-db-ignored-attributes)
 848                                 (if (find-charset name)
 849                                     (push name ccss))
 850                                 (push name atr-d)))
 851                             atr-d))
 852                     (dolist (name (char-attribute-list))
 853                       (unless (memq name char-db-ignored-attributes)
 854                         (if (find-charset name)
 855                             (push name ccss))
 856                         (push name atr-d)))
 857                     atr-d)
 858                   #'char-attribute-name<)))
 859     (insert "(")
 860     (when (memq '<-subsumptive attributes)
 861       (when (or readable (not for-sub-node))
 862         (when (setq value (get-char-attribute char '<-subsumptive))
 863           (char-db-insert-relation-feature char '<-subsumptive value
 864                                            line-breaking
 865                                            ccss readable)))
 866       (setq attributes (delq '<-subsumptive attributes)))
 867     (when (and (memq '<-denotational attributes)
 868                (setq value (get-char-attribute char '<-denotational)))
 869       (char-db-insert-relation-feature char '<-denotational value
 870                                        line-breaking
 871                                        ccss readable)
 872       (setq attributes (delq '<-denotational attributes)))
 873     (when (and (memq '<-denotational@component attributes)
 874                (setq value (get-char-attribute char '<-denotational@component)))
 875       (char-db-insert-relation-feature char '<-denotational@component value
 876                                        line-breaking
 877                                        ccss readable)
 878       (setq attributes (delq '<-denotational@component attributes)))
 879     (when (and (memq 'name attributes)
 880                (setq value (get-char-attribute char 'name)))
 881       (insert (format
 882                (if (> (+ (current-column) (length value)) 48)
 883                    "(name . %S)%s"
 884                  "(name               . %S)%s")
 885                value line-breaking))
 886       (setq attributes (delq 'name attributes))
 887       )
 888     (when (and (memq 'name* attributes)
 889                (setq value (get-char-attribute char 'name*)))
 890       (insert (format
 891                (if (> (+ (current-column) (length value)) 48)
 892                    "(name* . %S)%s"
 893                  "(name*              . %S)%s")
 894                value line-breaking))
 895       (setq attributes (delq 'name* attributes))
 896       )
 897     (when (and (memq 'script attributes)
 898                (setq value (get-char-attribute char 'script)))
 899       (insert (format "(script\t\t%s)%s"
 900                       (mapconcat (function prin1-to-string)
 901                                  value " ")
 902                       line-breaking))
 903       (setq attributes (delq 'script attributes))
 904       )
 905     (dolist (name '(=>ucs =>ucs*))
 906       (when (and (memq name attributes)
 907                  (setq value (get-char-attribute char name)))
 908         (insert (format "(%-18s . #x%04X)\t; %c%s"
 909                         name value (decode-char '=ucs value)
 910                         line-breaking))
 911         (setq attributes (delq name attributes))))
 912     (dolist (name '(=>ucs@gb =>ucs@big5))
 913       (when (and (memq name attributes)
 914                  (setq value (get-char-attribute char name)))
 915         (insert (format "(%-18s . #x%04X)\t; %c%s"
 916                         name value
 917                         (decode-char (intern
 918                                       (concat "="
 919                                               (substring
 920                                                (symbol-name name) 2)))
 921                                      value)
 922                         line-breaking))
 923         (setq attributes (delq name attributes))
 924         ))
 925     ;; (dolist (name '(=>daikanwa))
 926     ;;   (when (and (memq name attributes)
 927     ;;              (setq value (get-char-attribute char name)))
 928     ;;     (insert
 929     ;;      (if (integerp value)
 930     ;;          (format "(%-18s . %05d)\t; %c%s"
 931     ;;                  name value (decode-char '=daikanwa value)
 932     ;;                  line-breaking)
 933     ;;        (format "(%-18s %s)\t; %c%s"
 934     ;;                name
 935     ;;                (mapconcat (function prin1-to-string)
 936     ;;                           value " ")
 937     ;;                (char-representative-of-daikanwa char)
 938     ;;                line-breaking)))
 939     ;;     (setq attributes (delq name attributes))))
 940     (when (and (memq 'general-category attributes)
 941                (setq value (get-char-attribute char 'general-category)))
 942       (insert (format
 943                "(general-category\t%s) ; %s%s"
 944                (mapconcat (lambda (cell)
 945                             (format "%S" cell))
 946                           value " ")
 947                (cond ((rassoc value unidata-normative-category-alist)
 948                       "Normative Category")
 949                      ((rassoc value unidata-informative-category-alist)
 950                       "Informative Category")
 951                      (t
 952                       "Unknown Category"))
 953                line-breaking))
 954       (setq attributes (delq 'general-category attributes))
 955       )
 956     (when (and (memq 'bidi-category attributes)
 957                (setq value (get-char-attribute char 'bidi-category)))
 958       (insert (format "(bidi-category\t. %S)%s"
 959                       value
 960                       line-breaking))
 961       (setq attributes (delq 'bidi-category attributes))
 962       )
 963     (unless (or (not (memq 'mirrored attributes))
 964                 (eq (setq value (get-char-attribute char 'mirrored 'empty))
 965                     'empty))
 966       (insert (format "(mirrored\t\t. %S)%s"
 967                       value
 968                       line-breaking))
 969       (setq attributes (delq 'mirrored attributes))
 970       )
 971     (cond
 972      ((and (memq 'decimal-digit-value attributes)
 973            (setq value (get-char-attribute char 'decimal-digit-value)))
 974       (insert (format "(decimal-digit-value . %S)%s"
 975                       value
 976                       line-breaking))
 977       (setq attributes (delq 'decimal-digit-value attributes))
 978       (when (and (memq 'digit-value attributes)
 979                  (setq value (get-char-attribute char 'digit-value)))
 980         (insert (format "(digit-value\t . %S)%s"
 981                         value
 982                         line-breaking))
 983         (setq attributes (delq 'digit-value attributes))
 984         )
 985       (when (and (memq 'numeric-value attributes)
 986                  (setq value (get-char-attribute char 'numeric-value)))
 987         (insert (format "(numeric-value\t . %S)%s"
 988                         value
 989                         line-breaking))
 990         (setq attributes (delq 'numeric-value attributes))
 991         )
 992       )
 993      (t
 994       (when (and (memq 'digit-value attributes)
 995                  (setq value (get-char-attribute char 'digit-value)))
 996         (insert (format "(digit-value\t. %S)%s"
 997                         value
 998                         line-breaking))
 999         (setq attributes (delq 'digit-value attributes))
1000         )
1001       (when (and (memq 'numeric-value attributes)
1002                  (setq value (get-char-attribute char 'numeric-value)))
1003         (insert (format "(numeric-value\t. %S)%s"
1004                         value
1005                         line-breaking))
1006         (setq attributes (delq 'numeric-value attributes))
1007         )))
1008     (when (and (memq 'iso-10646-comment attributes)
1009                (setq value (get-char-attribute char 'iso-10646-comment)))
1010       (insert (format "(iso-10646-comment\t. %S)%s"
1011                       value
1012                       line-breaking))
1013       (setq attributes (delq 'iso-10646-comment attributes))
1014       )
1015     (when (and (memq 'morohashi-daikanwa attributes)
1016                (setq value (get-char-attribute char 'morohashi-daikanwa)))
1017       (insert (format "(morohashi-daikanwa\t%s)%s"
1018                       (mapconcat (function prin1-to-string) value " ")
1019                       line-breaking))
1020       (setq attributes (delq 'morohashi-daikanwa attributes))
1021       )
1022     (setq radical nil
1023           strokes nil)
1024     (when (and (memq 'ideographic-radical attributes)
1025                (setq value (get-char-attribute char 'ideographic-radical)))
1026       (setq radical value)
1027       (insert (format "(ideographic-radical . %S)\t; %c%s"
1028                       radical
1029                       (ideographic-radical radical)
1030                       line-breaking))
1031       (setq attributes (delq 'ideographic-radical attributes))
1032       )
1033     (when (and (memq 'shuowen-radical attributes)
1034                (setq value (get-char-attribute char 'shuowen-radical)))
1035       (insert (format "(shuowen-radical\t. %S)\t; %c%s"
1036                       value
1037                       (shuowen-radical value)
1038                       line-breaking))
1039       (setq attributes (delq 'shuowen-radical attributes))
1040       )
1041     (let (key)
1042       (dolist (domain
1043                (append
1044                 char-db-feature-domains
1045                 (let (dest domain)
1046                   (dolist (feature (char-attribute-list))
1047                     (setq feature (symbol-name feature))
1048                     (when (string-match
1049                            "\\(radical\\|strokes\\)@\\([^@*]+\\)\\(\\*\\|$\\)"
1050                            feature)
1051                       (setq domain (intern (match-string 2 feature)))
1052                      (unless (memq domain dest)
1053                        (setq dest (cons domain dest)))))
1054                   (sort dest #'string<))))
1055         (setq key (intern (format "%s@%s" 'ideographic-radical domain)))
1056         (when (and (memq key attributes)
1057                    (setq value (get-char-attribute char key)))
1058           (setq radical value)
1059           (insert (format "(%s . %S)\t; %c%s"
1060                           key
1061                           radical
1062                           (ideographic-radical radical)
1063                           line-breaking))
1064           (setq attributes (delq key attributes))
1065           )
1066         (setq key (intern (format "%s@%s" 'ideographic-strokes domain)))
1067         (when (and (memq key attributes)
1068                    (setq value (get-char-attribute char key)))
1069           (setq strokes value)
1070           (insert (format "(%s . %S)%s"
1071                           key
1072                           strokes
1073                           line-breaking))
1074           (setq attributes (delq key attributes))
1075           )
1076         (setq key (intern (format "%s@%s" 'total-strokes domain)))
1077         (when (and (memq key attributes)
1078                    (setq value (get-char-attribute char key)))
1079           (insert (format "(%s       . %S)%s"
1080                           key
1081                           value
1082                           line-breaking))
1083           (setq attributes (delq key attributes))
1084           )
1085         (dolist (feature '(ideographic-radical
1086                            ideographic-strokes
1087                            total-strokes))
1088           (setq key (intern (format "%s@%s*sources" feature domain)))
1089           (when (and (memq key attributes)
1090                      (setq value (get-char-attribute char key)))
1091             (insert (format "(%s%s" key line-breaking))
1092             (dolist (cell value)
1093               (insert (format " %s" cell)))
1094             (insert ")")
1095             (insert line-breaking)
1096             (setq attributes (delq key attributes))
1097             ))
1098         ))
1099     (when (and (memq 'ideographic-strokes attributes)
1100                (setq value (get-char-attribute char 'ideographic-strokes)))
1101       (setq strokes value)
1102       (insert (format "(ideographic-strokes . %S)%s"
1103                       strokes
1104                       line-breaking))
1105       (setq attributes (delq 'ideographic-strokes attributes))
1106       )
1107     (when (and (memq 'kangxi-radical attributes)
1108                (setq value (get-char-attribute char 'kangxi-radical)))
1109       (unless (eq value radical)
1110         (insert (format "(kangxi-radical\t . %S)\t; %c%s"
1111                         value
1112                         (ideographic-radical value)
1113                         line-breaking))
1114         (or radical
1115             (setq radical value)))
1116       (setq attributes (delq 'kangxi-radical attributes))
1117       )
1118     (when (and (memq 'kangxi-strokes attributes)
1119                (setq value (get-char-attribute char 'kangxi-strokes)))
1120       (unless (eq value strokes)
1121         (insert (format "(kangxi-strokes\t . %S)%s"
1122                         value
1123                         line-breaking))
1124         (or strokes
1125             (setq strokes value)))
1126       (setq attributes (delq 'kangxi-strokes attributes))
1127       )
1128     (when (and (memq 'japanese-radical attributes)
1129                (setq value (get-char-attribute char 'japanese-radical)))
1130       (unless (eq value radical)
1131         (insert (format "(japanese-radical\t . %S)\t; %c%s"
1132                         value
1133                         (ideographic-radical value)
1134                         line-breaking))
1135         (or radical
1136             (setq radical value)))
1137       (setq attributes (delq 'japanese-radical attributes))
1138       )
1139     (when (and (memq 'japanese-strokes attributes)
1140                (setq value (get-char-attribute char 'japanese-strokes)))
1141       (unless (eq value strokes)
1142         (insert (format "(japanese-strokes\t . %S)%s"
1143                         value
1144                         line-breaking))
1145         (or strokes
1146             (setq strokes value)))
1147       (setq attributes (delq 'japanese-strokes attributes))
1148       )
1149     (when (and (memq 'cns-radical attributes)
1150                (setq value (get-char-attribute char 'cns-radical)))
1151       (insert (format "(cns-radical\t . %S)\t; %c%s"
1152                       value
1153                       (ideographic-radical value)
1154                       line-breaking))
1155       (setq attributes (delq 'cns-radical attributes))
1156       )
1157     (when (and (memq 'cns-strokes attributes)
1158                (setq value (get-char-attribute char 'cns-strokes)))
1159       (unless (eq value strokes)
1160         (insert (format "(cns-strokes\t . %S)%s"
1161                         value
1162                         line-breaking))
1163         (or strokes
1164             (setq strokes value)))
1165       (setq attributes (delq 'cns-strokes attributes))
1166       )
1167     ;; (when (and (memq 'shinjigen-1-radical attributes)
1168     ;;            (setq value (get-char-attribute char 'shinjigen-1-radical)))
1169     ;;   (unless (eq value radical)
1170     ;;     (insert (format "(shinjigen-1-radical . %S)\t; %c%s"
1171     ;;                     value
1172     ;;                     (ideographic-radical value)
1173     ;;                     line-breaking))
1174     ;;     (or radical
1175     ;;         (setq radical value)))
1176     ;;   (setq attributes (delq 'shinjigen-1-radical attributes))
1177     ;;   )
1178     (when (and (memq 'ideographic- attributes)
1179                (setq value (get-char-attribute char 'ideographic-)))
1180       (insert "(ideographic-       ")
1181       (setq lbs (concat "\n" (make-string (current-column) ?\ ))
1182             separator nil)
1183       (while (consp value)
1184         (setq cell (car value))
1185         (if (integerp cell)
1186             (setq cell (decode-char '=ucs cell)))
1187         (cond ((characterp cell)
1188                (if separator
1189                    (insert lbs))
1190                (if readable
1191                    (insert (format "%S" cell))
1192                  (char-db-insert-char-spec cell readable))
1193                (setq separator lbs))
1194               ((consp cell)
1195                (if separator
1196                    (insert lbs))
1197                (if (consp (car cell))
1198                    (char-db-insert-char-spec cell readable)
1199                  (char-db-insert-char-reference cell readable))
1200                (setq separator lbs))
1201               (t
1202                (if separator
1203                    (insert separator))
1204                (insert (prin1-to-string cell))
1205                (setq separator " ")))
1206         (setq value (cdr value)))
1207       (insert ")")
1208       (insert line-breaking)
1209       (setq attributes (delq 'ideographic- attributes)))
1210     (when (and (memq 'total-strokes attributes)
1211                (setq value (get-char-attribute char 'total-strokes)))
1212       (insert (format "(total-strokes       . %S)%s"
1213                       value
1214                       line-breaking))
1215       (setq attributes (delq 'total-strokes attributes))
1216       )
1217     (when (and (memq '->ideograph attributes)
1218                (setq value (get-char-attribute char '->ideograph)))
1219       (insert (format "(->ideograph\t%s)%s"
1220                       (mapconcat (lambda (code)
1221                                    (cond ((symbolp code)
1222                                           (symbol-name code))
1223                                          ((integerp code)
1224                                           (format "#x%04X" code))
1225                                          (t
1226                                           (format "%s %S"
1227                                                   line-breaking code))))
1228                                  value " ")
1229                       line-breaking))
1230       (setq attributes (delq '->ideograph attributes))
1231       )
1232     ;; (when (and (memq '->decomposition attributes)
1233     ;;            (setq value (get-char-attribute char '->decomposition)))
1234     ;;   (insert (format "(->decomposition\t%s)%s"
1235     ;;                   (mapconcat (lambda (code)
1236     ;;                                (cond ((symbolp code)
1237     ;;                                       (symbol-name code))
1238     ;;                                      ((characterp code)
1239     ;;                                       (if readable
1240     ;;                                           (format "%S" code)
1241     ;;                                         (format "#x%04X"
1242     ;;                                                 (char-int code))
1243     ;;                                         ))
1244     ;;                                      ((integerp code)
1245     ;;                                       (format "#x%04X" code))
1246     ;;                                      (t
1247     ;;                                       (format "%s%S" line-breaking code))))
1248     ;;                              value " ")
1249     ;;                   line-breaking))
1250     ;;   (setq attributes (delq '->decomposition attributes))
1251     ;;   )
1252     (if (equal (get-char-attribute char '->titlecase)
1253                (get-char-attribute char '->uppercase))
1254         (setq attributes (delq '->titlecase attributes)))
1255     (when (and (memq '->mojikyo attributes)
1256                (setq value (get-char-attribute char '->mojikyo)))
1257       (insert (format "(->mojikyo\t\t. %06d)\t; %c%s"
1258                       value (decode-char 'mojikyo value)
1259                       line-breaking))
1260       (setq attributes (delq '->mojikyo attributes))
1261       )
1262     (when (and (memq 'hanyu-dazidian-vol attributes)
1263                (setq value (get-char-attribute char 'hanyu-dazidian-vol)))
1264       (insert (format "(hanyu-dazidian-vol  . %d)%s"
1265                       value line-breaking))
1266       (setq attributes (delq 'hanyu-dazidian-vol attributes))
1267       )
1268     (when (and (memq 'hanyu-dazidian-page attributes)
1269                (setq value (get-char-attribute char 'hanyu-dazidian-page)))
1270       (insert (format "(hanyu-dazidian-page . %d)%s"
1271                       value line-breaking))
1272       (setq attributes (delq 'hanyu-dazidian-page attributes))
1273       )
1274     (when (and (memq 'hanyu-dazidian-char attributes)
1275                (setq value (get-char-attribute char 'hanyu-dazidian-char)))
1276       (insert (format "(hanyu-dazidian-char . %d)%s"
1277                       value line-breaking))
1278       (setq attributes (delq 'hanyu-dazidian-char attributes))
1279       )
1280     (unless readable
1281       (dolist (ignored '(composition
1282                          ->denotational <-subsumptive ->ucs-unified
1283                          ->ideographic-component-forms))
1284         (setq attributes (delq ignored attributes))))
1285     (while attributes
1286       (setq name (car attributes))
1287       (unless (eq (setq value (get-char-attribute char name 'value-is-empty))
1288                   'value-is-empty)
1289         (cond ((setq ret (find-charset name))
1290                (setq name (charset-name ret))
1291                (when (not (memq name dest-ccss))
1292                  (setq dest-ccss (cons name dest-ccss))
1293                  (char-db-insert-ccs-feature name value line-breaking))
1294                )
1295               ((string-match "^=>ucs@" (symbol-name name))
1296                (insert (format "(%-18s . #x%04X)\t; %c%s"
1297                                name value (decode-char '=ucs value)
1298                                line-breaking))
1299                )
1300               ((eq name 'jisx0208-1978/4X)
1301                (insert (format "(%-18s . #x%04X)%s"
1302                                name value
1303                                line-breaking))
1304                )
1305               ((and
1306                 (not readable)
1307                 (not (eq name '->subsumptive))
1308                 (not (eq name '->uppercase))
1309                 (not (eq name '->lowercase))
1310                 (not (eq name '->titlecase))
1311                 (not (eq name '->canonical))
1312                 (not (eq name '->Bopomofo))
1313                 (not (eq name '->mistakable))
1314                 (not (eq name '->ideographic-variants))
1315                 (null (get-char-attribute
1316                        char (intern (format "%s*sources" name))))
1317                 (not (string-match "\\*sources$" (symbol-name name)))
1318                 (null (get-char-attribute
1319                        char (intern (format "%s*note" name))))
1320                 (not (string-match "\\*note$" (symbol-name name)))
1321                 (or (eq name '<-identical)
1322                     (eq name '<-uppercase)
1323                     (eq name '<-lowercase)
1324                     (eq name '<-titlecase)
1325                     (eq name '<-canonical)
1326                     (eq name '<-ideographic-variants)
1327                     ;; (eq name '<-synonyms)
1328                     (string-match "^<-synonyms" (symbol-name name))
1329                     (eq name '<-mistakable)
1330                     (when (string-match "^->" (symbol-name name))
1331                       (cond
1332                        ((string-match "^->fullwidth" (symbol-name name))
1333                         (not (and (consp value)
1334                                   (characterp (car value))
1335                                   (encode-char
1336                                    (car value) '=ucs 'defined-only)))
1337                         )
1338                        (t)))
1339                     ))
1340                )
1341               ((or (eq name 'ideographic-structure)
1342                    (eq name 'ideographic-combination)
1343                    (eq name 'ideographic-)
1344                    (eq name '=decomposition)
1345                    (char-feature-base-name= '=decomposition name)
1346                    (char-feature-base-name= '=>decomposition name)
1347                    ;; (string-match "^=>*decomposition\\(@[^*]+\\)?$"
1348                    ;;               (symbol-name name))
1349                    (string-match "^\\(->\\|<-\\)[^*]*$" (symbol-name name))
1350                    (string-match "^\\(->\\|<-\\)[^*]*\\*sources$"
1351                                  (symbol-name name))
1352                    )
1353                (char-db-insert-relation-feature char name value
1354                                                 line-breaking
1355                                                 ccss readable))
1356               ((memq name '(ideograph=
1357                             original-ideograph-of
1358                             ancient-ideograph-of
1359                             vulgar-ideograph-of
1360                             wrong-ideograph-of
1361                             ;; simplified-ideograph-of
1362                             ideographic-variants
1363                             ;; ideographic-different-form-of
1364                             ))
1365                (insert (format "(%-18s%s " name line-breaking))
1366                (setq lbs (concat "\n" (make-string (current-column) ?\ ))
1367                      separator nil)
1368                (while (consp value)
1369                  (setq cell (car value))
1370                  (if (and (consp cell)
1371                           (consp (car cell)))
1372                      (progn
1373                        (if separator
1374                            (insert lbs))
1375                        (char-db-insert-alist cell readable)
1376                        (setq separator lbs))
1377                    (if separator
1378                        (insert separator))
1379                    (insert (prin1-to-string cell))
1380                    (setq separator " "))
1381                  (setq value (cdr value)))
1382                (insert ")")
1383                (insert line-breaking))
1384               ((consp value)
1385                (insert (format "(%-18s " name))
1386                (setq lbs (concat "\n" (make-string (current-column) ?\ ))
1387                      separator nil)
1388                (while (consp value)
1389                  (setq cell (car value))
1390                  (if (and (consp cell)
1391                           (consp (car cell))
1392                           (setq ret (condition-case nil
1393                                         (find-char cell)
1394                                       (error nil))))
1395                      (progn
1396                        (setq rest cell
1397                              al nil
1398                              cal nil)
1399                        (while rest
1400                          (setq key (car (car rest)))
1401                          (if (find-charset key)
1402                              (setq cal (cons key cal))
1403                            (setq al (cons key al)))
1404                          (setq rest (cdr rest)))
1405                        (if separator
1406                            (insert lbs))
1407                        (insert-char-attributes ret
1408                                                readable
1409                                                al ; cal
1410                                                nil 'for-sub-node)
1411                        (setq separator lbs))
1412                    (setq ret (prin1-to-string cell))
1413                    (if separator
1414                        (if (< (+ (current-column)
1415                                  (length ret)
1416                                  (length separator))
1417                               76)
1418                            (insert separator)
1419                          (insert lbs)))
1420                    (insert ret)
1421                    (setq separator " "))
1422                  (setq value (cdr value)))
1423                (insert ")")
1424                (insert line-breaking))
1425               (t
1426                (insert (format "(%-18s" name))
1427                (setq ret (prin1-to-string value))
1428                (unless (< (+ (current-column)
1429                              (length ret)
1430                              3)
1431                           76)
1432                  (insert line-breaking))
1433                (insert " . " ret ")" line-breaking)
1434                ;; (insert (format "(%-18s . %S)%s"
1435                ;;                 name value
1436                ;;                 line-breaking))
1437                )
1438               ))
1439       (setq attributes (cdr attributes)))
1440     (insert ")")))
1441
1442 (defun insert-char-data (char &optional readable
1443                               attributes)
1444   (save-restriction
1445     (narrow-to-region (point)(point))
1446     (insert "(define-char
1447   '")
1448     (insert-char-attributes char readable attributes)
1449     (insert ")\n")
1450     (goto-char (point-min))
1451     (while (re-search-forward "[ \t]+$" nil t)
1452       (replace-match ""))
1453     ;; from tabify.
1454     (goto-char (point-min))
1455     (while (re-search-forward "[ \t][ \t][ \t]*" nil t)
1456       (let ((column (current-column))
1457             (indent-tabs-mode t))
1458         (delete-region (match-beginning 0) (point))
1459         (indent-to column)))
1460     (goto-char (point-max))
1461     ;; (tabify (point-min)(point-max))
1462     ))
1463
1464 (defun insert-char-data-with-variant (char &optional printable
1465                                            no-ucs-unified
1466                                            script excluded-script)
1467   (insert-char-data char printable)
1468   (let ((variants (char-variants char))
1469         rest
1470         variant vs ret)
1471     (setq variants (sort variants #'<))
1472     (setq rest variants)
1473     (setq variants (cons char variants))
1474     (while rest
1475       (setq variant (car rest))
1476       (unless (get-char-attribute variant '<-subsumptive)
1477         (if (and (or (null script)
1478                      (null (setq vs (get-char-attribute variant 'script)))
1479                      (memq script vs))
1480                  (or (null excluded-script)
1481                      (null (setq vs (get-char-attribute variant 'script)))
1482                      (not (memq excluded-script vs))))
1483             (unless (and no-ucs-unified (get-char-attribute variant '=ucs))
1484               (insert-char-data variant printable)
1485               (if (setq ret (char-variants variant))
1486                   (while ret
1487                     (or (memq (car ret) variants)
1488                         ;; (get-char-attribute (car ret) '<-subsumptive)
1489                         (setq rest (nconc rest (list (car ret)))))
1490                     (setq ret (cdr ret)))))))
1491       (setq rest (cdr rest)))))
1492
1493 (defun insert-char-range-data (min max &optional script excluded-script)
1494   (let ((code min)
1495         char)
1496     (while (<= code max)
1497       (setq char (decode-char '=ucs code))
1498       (if (encode-char char '=ucs 'defined-only)
1499           (insert-char-data-with-variant char nil 'no-ucs-unified
1500                                          script excluded-script))
1501       (setq code (1+ code)))))
1502
1503 (defun write-char-range-data-to-file (min max file
1504                                           &optional script excluded-script)
1505   (let ((coding-system-for-write char-db-file-coding-system))
1506     (with-temp-buffer
1507       (insert (format ";; -*- coding: %s -*-\n"
1508                       char-db-file-coding-system))
1509       (insert-char-range-data min max script excluded-script)
1510       (write-region (point-min)(point-max) file))))
1511
1512 (defvar what-character-original-window-configuration)
1513
1514 ;;;###autoload
1515 (defun what-char-definition (char)
1516   (interactive (list (char-after)))
1517   (let ((buf (get-buffer-create "*Character Description*"))
1518         (the-buf (current-buffer))
1519         (win-conf (current-window-configuration)))
1520     (pop-to-buffer buf)
1521     (make-local-variable 'what-character-original-window-configuration)
1522     (setq what-character-original-window-configuration win-conf)
1523     (setq buffer-read-only nil)
1524     (erase-buffer)
1525     (condition-case err
1526         (progn
1527           (insert-char-data-with-variant char 'printable)
1528           (unless (char-attribute-alist char)
1529             (insert (format ";; = %c\n"
1530                             (let* ((rest (split-char char))
1531                                    (ccs (pop rest))
1532                                    (code (pop rest)))
1533                               (while rest
1534                                 (setq code (logior (lsh code 8)
1535                                                    (pop rest))))
1536                               (decode-char ccs code)))))
1537           ;; (char-db-update-comment)
1538           (set-buffer-modified-p nil)
1539           (view-mode the-buf (lambda (buf)
1540                                (set-window-configuration
1541                                 what-character-original-window-configuration)
1542                                ))
1543           (goto-char (point-min)))
1544       (error (progn
1545                (set-window-configuration
1546                 what-character-original-window-configuration)
1547                (signal (car err) (cdr err)))))))
1548
1549
1550 ;;; @ end
1551 ;;;
1552
1553 (provide 'char-db-util)
1554
1555 ;;; char-db-util.el ends here