git.chise.org Git - chise/xemacs-chise.git.1/blob - lisp/utf-2000/char-db-util.el

   1 ;;; char-db-util.el --- Character Database utility -*- coding: utf-8-er; -*-
   2
   3 ;; Copyright (C) 1998,1999,2000,2001,2002,2003,2004,2005,2006,2007,
   4 ;;   2008,2009,2010,2011,2012,2013,2014,2015,2016,2017 MORIOKA Tomohiko.
   5
   6 ;; Author: MORIOKA Tomohiko <tomo@kanji.zinbun.kyoto-u.ac.jp>
   7 ;; Keywords: CHISE, Character Database, ISO/IEC 10646, UCS, Unicode, MULE.
   8
   9 ;; This file is part of XEmacs CHISE.
  10
  11 ;; XEmacs CHISE is free software; you can redistribute it and/or
  12 ;; modify it under the terms of the GNU General Public License as
  13 ;; published by the Free Software Foundation; either version 2, or (at
  14 ;; your option) any later version.
  15
  16 ;; XEmacs CHISE is distributed in the hope that it will be useful,
  17 ;; but WITHOUT ANY WARRANTY; without even the implied warranty of
  18 ;; MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the GNU
  19 ;; General Public License for more details.
  20
  21 ;; You should have received a copy of the GNU General Public License
  22 ;; along with XEmacs CHISE; see the file COPYING.  If not, write to
  23 ;; the Free Software Foundation, Inc., 59 Temple Place - Suite 330,
  24 ;; Boston, MA 02111-1307, USA.
  25
  26 ;;; Code:
  27
  28 (require 'chise-subr)
  29 (require 'ideograph-subr)
  30
  31 (defconst unidata-normative-category-alist
  32   '(("Lu" letter        uppercase)
  33     ("Ll" letter        lowercase)
  34     ("Lt" letter        titlecase)
  35     ("Mn" mark          non-spacing)
  36     ("Mc" mark          spacing-combining)
  37     ("Me" mark          enclosing)
  38     ("Nd" number        decimal-digit)
  39     ("Nl" number        letter)
  40     ("No" number        other)
  41     ("Zs" separator     space)
  42     ("Zl" separator     line)
  43     ("Zp" separator     paragraph)
  44     ("Cc" other         control)
  45     ("Cf" other         format)
  46     ("Cs" other         surrogate)
  47     ("Co" other         private-use)
  48     ("Cn" other         not-assigned)))
  49
  50 (defconst unidata-informative-category-alist
  51   '(("Lm" letter        modifier)
  52     ("Lo" letter        other)
  53     ("Pc" punctuation   connector)
  54     ("Pd" punctuation   dash)
  55     ("Ps" punctuation   open)
  56     ("Pe" punctuation   close)
  57     ("Pi" punctuation   initial-quote)
  58     ("Pf" punctuation   final-quote)
  59     ("Po" punctuation   other)
  60     ("Sm" symbol        math)
  61     ("Sc" symbol        currency)
  62     ("Sk" symbol        modifier)
  63     ("So" symbol        other)
  64     ))
  65
  66 (defconst shuowen-radicals
  67   [?一 ?上 ?示 ?三 ?王 ?玉 ?玨 ?气 ?士 ?丨 ; 010
  68    ?屮 ?艸 ?蓐 ?茻 ?小 ?八 ?釆 ?半 ?牛 ?犛 ; 020
  69    ?告 ?口 ?凵 ?吅 ?哭 ?走 ?止 ?癶 ?步 ?此 ; 030
  70    ?正 ?是 ?辵 ?彳 ?廴 ?㢟 ?行 ?齒 ?牙 ?足 ; 040
  71    ?疋 ?品 ?龠 ?冊 ?㗊 ?舌 ?干 ?谷 ?只 ?㕯 ; 050
  72    ?句 ?丩 ?古 ?十 ?卅 ?言 ?誩 ?音 ?䇂 ?丵 ; 060
  73    ?菐 ?𠬞 ?𠬜 ?共 ?異 ?舁 ?𦥑 ?䢅 ?爨 ?革 ; 070
  74    ?鬲 ?䰜 ?爪 ?𠃨 ?鬥 ?又 ?𠂇 ?㕜 ?支 ?𦘒 ; 080
  75    ?聿 ?畫 ?隶 ?臤 ?臣 ?殳 ?殺 ?𠘧 ?寸 ?皮 ; 090
  76    ?㼱 ?攴 ?敎 ?卜 ?用 ?爻 ?㸚 ?𥄎 ?目 ?䀠 ; 100
  77    ?眉 ?盾 ?自 ?白 ?鼻 ?皕 ?習 ?羽 ?隹 ?奞 ; 110
  78    ?萑 ?𦫳 ?苜 ?羊 ?羴 ?瞿 ?雔 ?雥 ?鳥 ?烏 ; 120
  79    ?𠦒 ?冓 ?幺 ?𢆶 ?叀 ?玄 ?予 ?放 ?𠬪 ?𣦼 ; 130
  80    ?歺 ?死 ?冎 ?骨 ?肉 ?筋 ?刀 ?刃 ?㓞 ?丰 ; 140
  81    ?耒 ?𧢲 ?竹 ?箕 ?丌 ?左 ?工 ?㠭 ?巫 ?甘 ; 150
  82    ?曰 ?乃 ?丂 ?可 ?兮 ?号 ?亏 ?旨 ?喜 ?壴 ; 160
  83 ;  ?旨 ?曰 ?乃 ?丂 ?可 ?兮 ?号 ?亏 ?喜 ?壴 ; 160
  84    ?鼓 ?豈 ?豆 ?豊 ?豐 ?䖒 ?虍 ?虎 ?虤 ?皿 ; 170
  85    ?𠙴 ?去 ?血 ?丶 ?丹 ?青 ?井 ?皀 ?鬯 ?食 ; 180
  86    ?亼 ?會 ?倉 ?入 ?缶 ?矢 ?高 ?冂 ?𩫏 ?京 ; 190
  87    ?亯 ?𣆪 ?畗 ?㐭 ?嗇 ?來 ?麥 ?夊 ?舛 ?䑞 ; 200
  88    ?韋 ?弟 ?夂 ?久 ?桀 ?木 ?東 ?林 ?才 ?叒 ; 210
  89    ?之 ?帀 ?出 ?𣎵 ?生 ?乇 ?𠂹 ?𠌶 ?華 ?𥝌 ; 220
  90    ?稽 ?巢 ?桼 ?束 ?㯻 ?囗 ?員 ?貝 ?邑 ?𨛜 ; 230
  91    ?日 ?旦 ?倝 ?㫃 ?冥 ?晶 ?月 ?有 ?明 ?囧 ; 240
  92    ?夕 ?多 ?毌 ?𢎘 ?𣐺 ?卣 ?齊 ?朿 ?片 ?鼎 ; 250
  93    ?克 ?彔 ?禾 ?秝 ?黍 ?香 ?米 ?毇 ?臼 ?凶 ; 260
  94    ?𣎳 ?林 ?麻 ?尗 ?耑 ?韭 ?瓜 ?瓠 ?宀 ?宮 ; 270
  95    ?呂 ?穴 ?㝱 ?𤕫 ?冖 ?𠔼 ?冃 ?㒳 ?网 ?襾 ; 280
  96    ?巾 ?巿 ?帛 ?白 ?㡀 ?黹 ?人 ?𠤎 ?匕 ?从 ; 290
  97    ?比 ?北 ?丘 ?㐺 ?𡈼 ?重 ?臥 ?身 ?㐆 ?衣 ; 300
  98    ?裘 ?老 ?毛 ?毳 ?尸 ?尺 ?尾 ?履 ?舟 ?方 ; 310
  99    ?儿 ?兄 ?兂 ?皃 ?𠑹 ?先 ?秃 ?見 ?覞 ?欠 ; 320
 100    ?㱃 ?㳄 ?旡 ?頁 ?𦣻 ?面 ?丏 ?首 ?𥄉 ?須 ; 330
 101    ?彡 ?彣 ?文 ?髟 ?后 ?司 ?卮 ?卩 ?印 ?色 ; 340
 102    ?𠨍 ?辟 ?勹 ?包 ?茍 ?鬼 ?甶 ?厶 ?嵬 ?山 ; 350
 103    ?屾 ?屵 ?广 ?厂 ?丸 ?危 ?石 ?長 ?勿 ?冄 ; 360
 104    ?而 ?豕 ?㣇 ?彑 ?豚 ?豸 ?𤉡 ?易 ?象 ?馬 ; 370
 105    ?𢊁 ?鹿 ?麤 ?㲋 ?兔 ?萈 ?犬 ?㹜 ?鼠 ?能 ; 380
 106    ?熊 ?火 ?炎 ?黑 ?囪 ?焱 ?炙 ?赤 ?大 ?亦 ; 390
 107    ])
 108
 109 (defun shuowen-radical (number)
 110   (aref shuowen-radicals (1- number)))
 111
 112 (defvar char-db-file-coding-system 'utf-8-mcs-er)
 113
 114 (defvar char-db-ignored-attributes '(ideographic-products))
 115
 116 (defvar char-db-coded-charset-priority-list
 117   '(ascii
 118     control-1
 119     latin-iso8859-1
 120     latin-iso8859-2
 121     latin-iso8859-3
 122     latin-iso8859-4
 123     latin-iso8859-9
 124     latin-jisx0201
 125     cyrillic-iso8859-5
 126     greek-iso8859-7
 127     thai-tis620
 128     ;; =mj
 129     ;; =adobe-japan1-0
 130     ;; =adobe-japan1-1
 131     ;; =adobe-japan1-2
 132     ;; =adobe-japan1-3
 133     ;; =adobe-japan1-4
 134     ;; =adobe-japan1-5
 135     ;; =adobe-japan1-6
 136     =jis-x0208
 137     =jis-x0208@1978
 138     =jis-x0208@1983
 139     japanese-jisx0212
 140     chinese-gb2312
 141     =jis-x0208@1990
 142     chinese-cns11643-1
 143     chinese-cns11643-2
 144     chinese-cns11643-3
 145     chinese-cns11643-4
 146     chinese-cns11643-5
 147     chinese-cns11643-6
 148     chinese-cns11643-7
 149     =jis-x0213-1
 150     =jis-x0213-1@2000
 151     =jis-x0213-1@2004
 152     =jis-x0213-2
 153     korean-ksc5601
 154     chinese-isoir165
 155     katakana-jisx0201
 156     hebrew-iso8859-8
 157     chinese-gb12345
 158     latin-viscii
 159     ethiopic-ucs
 160     =big5-cdp
 161     =hanyo-denshi/ja
 162     =hanyo-denshi/jb
 163     =hanyo-denshi/jc
 164     =hanyo-denshi/jd
 165     =hanyo-denshi/ft
 166     =hanyo-denshi/ia
 167     =hanyo-denshi/ib
 168     =hanyo-denshi/hg
 169     =hanyo-denshi/jt
 170     =hanyo-denshi/ks
 171     =hanyo-denshi/tk
 172     ;; ==mj
 173     ;; ==adobe-japan1-0
 174     ;; ==adobe-japan1-1
 175     ;; ==adobe-japan1-2
 176     ;; ==adobe-japan1-3
 177     ;; ==adobe-japan1-4
 178     ;; ==adobe-japan1-5
 179     ;; ==adobe-japan1-6
 180     ==jis-x0208
 181     ==jis-x0213-1
 182     ==jis-x0213-2
 183     ==jis-x0212
 184     ==hanyo-denshi/ja
 185     ==hanyo-denshi/jb
 186     ==hanyo-denshi/jc
 187     ==hanyo-denshi/ft
 188     ==hanyo-denshi/ia
 189     ==hanyo-denshi/ib
 190     ==hanyo-denshi/hg
 191     ==hanyo-denshi/jt
 192     ==hanyo-denshi/ks
 193     ==hanyo-denshi/tk
 194     =gt
 195     =gt-k
 196     =daikanwa
 197     =daikanwa@rev2
 198     =daikanwa@rev1
 199     =daikanwa/+p
 200     ==daikanwa
 201     =cbeta
 202     ideograph-hanziku-1
 203     ideograph-hanziku-2
 204     ideograph-hanziku-3
 205     ideograph-hanziku-4
 206     ideograph-hanziku-5
 207     ideograph-hanziku-6
 208     ideograph-hanziku-7
 209     ideograph-hanziku-8
 210     ideograph-hanziku-9
 211     ideograph-hanziku-10
 212     ideograph-hanziku-11
 213     ideograph-hanziku-12
 214     ;; =>>>adobe-japan1-0
 215     ;; =>>>adobe-japan1-1
 216     ;; =>>>adobe-japan1-2
 217     ;; =>>>adobe-japan1-3
 218     ;; =>>>adobe-japan1-4
 219     ;; =>>>adobe-japan1-5
 220     ;; =>>>adobe-japan1-6
 221     ;; =>>>jis-x0208
 222     ;; =>>>jis-x0213-1
 223     ;; =>>>jis-x0213-2
 224     ;; =>>>hanyo-denshi/ja
 225     ;; =>>>hanyo-denshi/jb
 226     ;; =>>>hanyo-denshi/jc
 227     ;; =>>>hanyo-denshi/ft
 228     ;; =>>>hanyo-denshi/ib
 229     ;; =>>>hanyo-denshi/hg
 230     ;; =>>>hanyo-denshi/jt
 231     ;; =>>>hanyo-denshi/ks
 232     ;; =>>>gt
 233     =>>adobe-japan1-0
 234     =>>adobe-japan1-1
 235     =>>adobe-japan1-2
 236     =>>adobe-japan1-3
 237     =>>adobe-japan1-4
 238     =>>adobe-japan1-5
 239     =>>adobe-japan1-6
 240     =>>jis-x0208
 241     =>>jis-x0213-1
 242     =>>jis-x0213-1@2000
 243     =>>jis-x0213-1@2004
 244     =>>jis-x0213-2
 245     =>>jis-x0208@1978
 246     =>>hanyo-denshi/ft
 247     =>>hanyo-denshi/jt
 248     =>>hanyo-denshi/ks
 249     =>>gt
 250     =>>daikanwa
 251     =>>cbeta
 252     =+>jis-x0208
 253     =+>jis-x0213-1
 254     =+>jis-x0213-2
 255     =+>adobe-japan1-0
 256     =+>adobe-japan1-1
 257     =+>adobe-japan1-2
 258     =+>adobe-japan1-3
 259     =+>adobe-japan1-4
 260     =+>adobe-japan1-5
 261     =+>adobe-japan1-6
 262     =+>jis-x0208@1978
 263     =>jis-x0208
 264     =>jis-x0208@1997
 265     =>jis-x0213-1
 266     =>jis-x0213-1@2000
 267     =>jis-x0213-1@2004
 268     =>jis-x0213-2
 269     ==>ucs@bucs
 270     =>iwds-1
 271     =>ucs@hanyo-denshi
 272     =>ucs@iso
 273     =>ucs@unicode
 274     =>ucs@jis
 275     =>ucs@cns
 276     =>ucs@ks
 277     =+>ucs@iso
 278     =+>ucs@unicode
 279     =+>ucs@jis
 280     =+>ucs@jis/1990
 281     =+>ucs@cns
 282     =+>ucs@ks
 283     =>>ucs@iso
 284     =>>ucs@unicode
 285     =>>ucs@jis
 286     =>>ucs@cns
 287     =>>>ucs@iso
 288     =>>>ucs@unicode
 289     ==ucs@iso
 290     ==ucs@unicode
 291     ;; ==ucs@cns
 292     ==gb2312
 293     ==ks-x1001
 294     ==cns11643-1
 295     ==cns11643-2
 296     ==cns11643-3
 297     ==cns11643-4
 298     ==cns11643-5
 299     ==cns11643-6
 300     ==cns11643-7
 301     ==gt
 302     ==jis-x0208@1990
 303     ;; ==jis-x0208@1983
 304     ==jis-x0208@1978
 305     ==gt-k
 306     =ucs@iso
 307     =ucs@unicode
 308     =ucs@cns
 309     ==big5-cdp
 310     ==cbeta
 311     =>>big5-cdp
 312     =>>gt-k
 313     =+>gt
 314     =>gt
 315     =>mj
 316     =>big5-cdp
 317     =>daikanwa
 318     =>daikanwa/ho
 319     =>cns11643-5
 320     =>cns11643-7
 321     =big5
 322     =big5-eten
 323     =>gt-k
 324     =zinbun-oracle
 325     =>zinbun-oracle
 326     =ruimoku-v6
 327     =>>ruimoku-v6
 328     ==ruimoku-v6
 329     =jef-china3
 330     =>cbeta
 331     =shinjigen
 332     =ucs-var-001
 333     =ucs-var-002
 334     =ucs-var-003
 335     =ucs-var-004
 336     =ucs-var-005
 337     =ucs-var-006
 338     =ucs-var-008
 339     =ucs-var-010
 340     =ucs-itaiji-001
 341     =ucs-itaiji-002
 342     =ucs-itaiji-003
 343     =ucs-itaiji-004
 344     =ucs-itaiji-005
 345     =ucs-itaiji-006
 346     =ucs-itaiji-007
 347     =ucs-itaiji-008
 348     =ucs-itaiji-009
 349     =ucs-itaiji-011
 350     =>ucs-itaiji-001
 351     =>ucs-itaiji-002
 352     =>ucs-itaiji-003
 353     =>ucs-itaiji-005
 354     =>ucs-itaiji-006
 355     =>ucs-itaiji-007
 356     =big5-cdp-var-001
 357     =big5-cdp-var-002
 358     =big5-cdp-var-003
 359     =big5-cdp-var-004
 360     =big5-cdp-var-005
 361     =big5-cdp-itaiji-001
 362     =big5-cdp-itaiji-002
 363     =big5-cdp-itaiji-003
 364     =>big5-cdp-itaiji-001
 365     =>ucs@iwds-1
 366     =>ucs@component
 367     ==ucs@gb
 368     =ucs@JP/hanazono))
 369
 370
 371 ;;; @ char-db formatters
 372 ;;;
 373
 374 (defun char-db-make-char-spec (char)
 375   (let (ret char-spec)
 376     (cond ((characterp char)
 377            (cond ((and (setq ret (encode-char char '=ucs 'defined-only))
 378                        (not (and (<= #xE000 ret)(<= ret #xF8FF))))
 379                   (setq char-spec (list (cons '=ucs ret)))
 380                   (cond ((setq ret (get-char-attribute char 'name))
 381                          (setq char-spec (cons (cons 'name ret) char-spec))
 382                          )
 383                         ((setq ret (get-char-attribute char 'name*))
 384                          (setq char-spec (cons (cons 'name* ret) char-spec))
 385                          ))
 386                   )
 387                  ((encode-char char '=mj 'defined-only)
 388                   (setq char-spec nil)
 389                   (dolist (ccs (charset-list))
 390                     (if (and (or (eq ccs '=mj)
 391                                  ;; (eq (charset-property ccs 'iso-ir) 177)
 392                                  (string-match "=ucs@" (symbol-name ccs))
 393                                  )
 394                              (setq ccs (charset-name ccs))
 395                              (null (assq ccs char-spec))
 396                              (setq ret (encode-char char ccs 'defined-only)))
 397                         (setq char-spec (cons (cons ccs ret) char-spec))))
 398                   )
 399                  ((encode-char char '==mj 'defined-only)
 400                   (setq char-spec nil)
 401                   (dolist (ccs (charset-list))
 402                     (if (and (or (eq ccs '==mj)
 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                  ((encode-char char '=adobe-japan1 'defined-only)
 412                   (setq char-spec nil)
 413                   (dolist (ccs (charset-list))
 414                     (if (and (or (memq ccs
 415                                        '(=adobe-japan1-0
 416                                          =adobe-japan1-1
 417                                          =adobe-japan1-2
 418                                          =adobe-japan1-3
 419                                          =adobe-japan1-4
 420                                          =adobe-japan1-5
 421                                          =adobe-japan1-6
 422                                          ))
 423                                  ;; (eq (charset-property ccs 'iso-ir) 177)
 424                                  (string-match "=ucs@" (symbol-name ccs))
 425                                  )
 426                              (setq ccs (charset-name ccs))
 427                              (null (assq ccs char-spec))
 428                              (setq ret (encode-char char ccs 'defined-only)))
 429                         (setq char-spec (cons (cons ccs ret) char-spec))))
 430                   )
 431                  ((encode-char char '==adobe-japan1 'defined-only)
 432                   (setq char-spec nil)
 433                   (dolist (ccs (charset-list))
 434                     (if (and (or (memq ccs
 435                                        '(==adobe-japan1-0
 436                                          ==adobe-japan1-1
 437                                          ==adobe-japan1-2
 438                                          ==adobe-japan1-3
 439                                          ==adobe-japan1-4
 440                                          ==adobe-japan1-5
 441                                          ==adobe-japan1-6
 442                                          ))
 443                                  ;; (eq (charset-property ccs 'iso-ir) 177)
 444                                  (string-match "=ucs@" (symbol-name ccs))
 445                                  )
 446                              (setq ccs (charset-name ccs))
 447                              (null (assq ccs char-spec))
 448                              (setq ret (encode-char char ccs 'defined-only)))
 449                         (setq char-spec (cons (cons ccs ret) char-spec))))
 450                   )
 451                  ((setq ret
 452                         (catch 'tag
 453                           (let ((rest char-db-coded-charset-priority-list)
 454                                 ccs)
 455                             (while rest
 456                               (setq ccs (charset-name
 457                                          (find-charset (car rest))))
 458                               (if (setq ret
 459                                         (encode-char char ccs
 460                                                      'defined-only))
 461                                   (throw 'tag (cons ccs ret)))
 462                               (setq rest (cdr rest))))))
 463                   (setq char-spec (list ret))
 464                   (dolist (ccs (delq (car ret) (charset-list)))
 465                     (if (and (or (charset-iso-final-char ccs)
 466                                  (memq ccs
 467                                        '(=daikanwa
 468                                          =daikanwa@rev2
 469                                          ;; =gt-k
 470                                          =jis-x0208@1997
 471                                          ))
 472                                  (eq (charset-property ccs 'iso-ir) 177)
 473                                  ;; (string-match "=ucs@" (symbol-name ccs))
 474                                  )
 475                              (setq ccs (charset-name ccs))
 476                              (null (assq ccs char-spec))
 477                              (setq ret (encode-char char ccs 'defined-only)))
 478                         (setq char-spec (cons (cons ccs ret) char-spec))))
 479                   (if (null char-spec)
 480                       (setq char-spec (split-char char)))
 481                   (cond ((setq ret (get-char-attribute char 'name))
 482                          (setq char-spec (cons (cons 'name ret) char-spec))
 483                          )
 484                         ((setq ret (get-char-attribute char 'name*))
 485                          (setq char-spec (cons (cons 'name* ret) char-spec))
 486                          ))
 487                   )
 488                  ((setq ret (get-char-attribute
 489                              char 'ideographic-combination))
 490                   (setq char-spec
 491                         (cons (cons 'ideographic-combination ret)
 492                               char-spec))
 493                   ))
 494            char-spec)
 495           ((consp char)
 496            char))))
 497
 498 (defun char-db-insert-char-spec (char &optional readable column
 499                                       required-features)
 500   (unless column
 501     (setq column (current-column)))
 502   (let (char-spec temp-char)
 503     (setq char-spec (char-db-make-char-spec char))
 504     (unless (or (characterp char) ; char
 505                 (condition-case nil
 506                     (setq char (find-char char-spec))
 507                   (error nil)))
 508       ;; define temporary character
 509       ;;   Current implementation is dirty.
 510       (setq temp-char (define-char (cons '(ideograph-daikanwa . 0)
 511                                          char-spec)))
 512       (remove-char-attribute temp-char 'ideograph-daikanwa)
 513       (setq char temp-char))
 514     (insert-char-attributes char
 515                             readable
 516                             (union (mapcar #'car char-spec)
 517                                    required-features)
 518                             nil 'for-sub-node)
 519     (when temp-char
 520       ;; undefine temporary character
 521       ;;   Current implementation is dirty.
 522       (setq char-spec (char-attribute-alist temp-char))
 523       (while char-spec
 524         (remove-char-attribute temp-char (car (car char-spec)))
 525         (setq char-spec (cdr char-spec))))))
 526
 527 (defun char-db-insert-alist (alist &optional readable column)
 528   (unless column
 529     (setq column (current-column)))
 530   (let ((line-breaking
 531          (concat "\n" (make-string (1+ column) ?\ )))
 532         name value
 533         ret al ; cal
 534         key
 535         lbs cell rest separator)
 536     (insert "(")
 537     (while alist
 538       (setq name (car (car alist))
 539             value (cdr (car alist)))
 540       (cond ((eq name 'char)
 541              (insert "(char . ")
 542              (if (setq ret (condition-case nil
 543                                (find-char value)
 544                              (error nil)))
 545                  (progn
 546                    (setq al nil
 547                          ;; cal nil
 548                          )
 549                    (while value
 550                      (setq key (car (car value)))
 551                      ;; (if (find-charset key)
 552                      ;;     (setq cal (cons key cal))
 553                      (setq al (cons key al))
 554                      ;; )
 555                      (setq value (cdr value)))
 556                    (insert-char-attributes ret
 557                                            readable
 558                                            (or al 'none) ; cal
 559                                            nil 'for-sub-node))
 560                (insert (prin1-to-string value)))
 561              (insert ")")
 562              (insert line-breaking))
 563             ((consp value)
 564              (insert (format "(%-18s " name))
 565              (setq lbs (concat "\n" (make-string (current-column) ?\ )))
 566              (while (consp value)
 567                (setq cell (car value))
 568                (if (and (consp cell)
 569                         (consp (car cell))
 570                         (setq ret (condition-case nil
 571                                       (find-char cell)
 572                                     (error nil)))
 573                         )
 574                    (progn
 575                      (setq rest cell
 576                            al nil
 577                            ;; cal nil
 578                            )
 579                      (while rest
 580                        (setq key (car (car rest)))
 581                        ;; (if (find-charset key)
 582                        ;;     (setq cal (cons key cal))
 583                        (setq al (cons key al))
 584                        ;; )
 585                        (setq rest (cdr rest)))
 586                      (if separator
 587                          (insert lbs))
 588                      (insert-char-attributes ret
 589                                              readable
 590                                              al ; cal
 591                                              nil 'for-sub-node)
 592                      (setq separator lbs))
 593                  (if separator
 594                      (insert separator))
 595                  (insert (prin1-to-string cell))
 596                  (setq separator " "))
 597                (setq value (cdr value)))
 598              (insert ")")
 599              (insert line-breaking))
 600             (t
 601              (insert (format "(%-18s . %S)%s"
 602                              name value
 603                              line-breaking))))
 604       (setq alist (cdr alist))))
 605   (insert ")"))
 606
 607 (defun char-db-insert-char-reference (plist &optional readable column)
 608   (unless column
 609     (setq column (current-column)))
 610   (let ((line-breaking
 611          (concat "\n" (make-string (1+ column) ?\ )))
 612         (separator "")
 613         name value)
 614     (insert "(")
 615     (while plist
 616       (setq name (pop plist))
 617       (setq value (pop plist))
 618       (cond ((eq name :char)
 619              (insert separator)
 620              (insert ":char\t")
 621              (cond ((numberp value)
 622                     (setq value (decode-char '=ucs value)))
 623                    ;; ((consp value)
 624                    ;;  (setq value (or (find-char value)
 625                    ;;                  value)))
 626                    )
 627              (char-db-insert-char-spec value readable)
 628              (insert line-breaking)
 629              (setq separator ""))
 630             ((eq name :radical)
 631              (insert (format "%s%s\t%d ; %c%s"
 632                              separator
 633                              name value
 634                              (ideographic-radical value)
 635                              line-breaking))
 636              (setq separator ""))
 637             (t
 638              (insert (format "%s%s\t%S" separator name value))
 639              (setq separator line-breaking)))
 640       ))
 641   (insert ")"))
 642
 643 (defun char-db-decode-isolated-char (ccs code-point)
 644   (let (ret)
 645     (setq ret
 646           (cond ((eq ccs 'arabic-iso8859-6)
 647                  (decode-char ccs code-point))
 648                 ;; ((eq ccs '=gt)
 649                 ;;  (decode-builtin-char '==gt code-point))
 650                 ((and (memq ccs '(=gt-pj-1
 651                                   =gt-pj-2
 652                                   =gt-pj-3
 653                                   =gt-pj-4
 654                                   =gt-pj-5
 655                                   =gt-pj-6
 656                                   =gt-pj-7
 657                                   =gt-pj-8
 658                                   =gt-pj-9
 659                                   =gt-pj-10
 660                                   =gt-pj-11))
 661                       (setq ret (decode-char ccs code-point))
 662                       (setq ret (encode-char ret '=gt 'defined-only)))
 663                  (decode-builtin-char '=gt ret))
 664                 (t
 665                  (decode-builtin-char ccs code-point))))
 666     (cond ((null ret)
 667            (or (decode-char ccs code-point)
 668                (define-char (list (cons ccs code-point)))))
 669           ((and (<= 0 (char-int ret))
 670                 (<= (char-int ret) #x1F))
 671            (decode-char '=ucs (+ #x2400 (char-int ret))))
 672           ((= (char-int ret) #x7F)
 673            ?\u2421)
 674           (t ret))))
 675
 676 (defvar char-db-convert-obsolete-format t)
 677
 678 (defun char-db-insert-ccs-feature (name value line-breaking)
 679   (cond
 680    ((integerp value)
 681     (insert
 682      (format
 683       (cond
 684        ((memq name '(=>iwds-1
 685                      ==shinjigen
 686                      =shinjigen
 687                      =shinjigen@1ed ==shinjigen@1ed
 688                      =shinjigen@rev ==shinjigen@rev
 689                      =shinjigen/+p@rev ==shinjigen/+p@rev
 690                      ===daikanwa/ho ==daikanwa/ho
 691                      =daikanwa/ho =>>daikanwa/ho =>daikanwa/ho))
 692         "(%-18s .  %04d)\t; %c")
 693        ((eq name '=shinjigen@1ed/24pr)
 694         "(%-18s . %04d)\t; %c")
 695        ((or
 696          (memq name
 697                '(===daikanwa
 698                  ==daikanwa =daikanwa =>>daikanwa =>daikanwa
 699                  =daikanwa@rev1 =daikanwa@rev2
 700                  =daikanwa/+p ==daikanwa/+p ===daikanwa/+p
 701                  =>>daikanwa/+p
 702                  =daikanwa/+2p =>>daikanwa/+2p
 703                  =gt ==gt ===gt
 704                  =>>gt =+>gt =>gt
 705                  =gt-k ==gt-k ===gt-k
 706                  =>>gt-k =>gt-k
 707                  =adobe-japan1-0 ==adobe-japan1-0 ===adobe-japan1-0
 708                  =adobe-japan1-1 ==adobe-japan1-1 ===adobe-japan1-1
 709                  =adobe-japan1-2 ==adobe-japan1-2 ===adobe-japan1-2
 710                  =adobe-japan1-3 ==adobe-japan1-3 ===adobe-japan1-3
 711                  =adobe-japan1-4 ==adobe-japan1-4 ===adobe-japan1-4
 712                  =adobe-japan1-5 ==adobe-japan1-5 ===adobe-japan1-5
 713                  =adobe-japan1-6 ==adobe-japan1-6 ===adobe-japan1-6
 714                  =>>adobe-japan1-0 =+>adobe-japan1-0
 715                  =>>adobe-japan1-1 =+>adobe-japan1-1
 716                  =>>adobe-japan1-2 =+>adobe-japan1-2
 717                  =>>adobe-japan1-3 =+>adobe-japan1-3
 718                  =>>adobe-japan1-4 =+>adobe-japan1-4
 719                  =>>adobe-japan1-5 =+>adobe-japan1-5
 720                  =>>adobe-japan1-6 =+>adobe-japan1-6
 721                  =>cbeta =cbeta =>>cbeta ==cbeta ===cbeta
 722                  =zinbun-oracle =>zinbun-oracle
 723                  ===hng-jou ===hng-keg ===hng-dng ===hng-mam
 724                  ===hng-drt ===hng-kgk ===hng-myz ===hng-kda
 725                  ===hng-khi ===hng-khm ===hng-hok ===hng-kyd ===hng-sok
 726                  ===hng-yhk ===hng-kak ===hng-kar ===hng-kae
 727                  ===hng-sys ===hng-tsu ===hng-tzj
 728                  ===hng-hos ===hng-nak ===hng-jhk
 729                  ===hng-hod ===hng-gok ===hng-ink ===hng-nto
 730                  ===hng-nkm ===hng-k24 ===hng-nkk
 731                  ===hng-kcc ===hng-kcj ===hng-kbk ===hng-sik
 732                  ===hng-skk ===hng-kyu ===hng-ksk ===hng-wan
 733                  ===hng-okd ===hng-wad ===hng-kmi
 734                  ===hng-zkd ===hng-doh ===hng-jyu
 735                  ===hng-tzs ===hng-kss ===hng-kyo
 736                  ===hng-smk))
 737          ;; (string-match "^=adobe-" (symbol-name name))
 738          )
 739         "(%-18s . %05d)\t; %c")
 740        ((memq name '(=hanyo-denshi/ks
 741                      ==hanyo-denshi/ks ===hanyo-denshi/ks
 742                      =>>hanyo-denshi/ks
 743                      =koseki ==koseki
 744                      =mj ==mj ===mj =>>mj =>mj
 745                      =zihai mojikyo))
 746         "(%-18s . %06d)\t; %c")
 747        ((memq name '(=hanyo-denshi/tk ==hanyo-denshi/tk))
 748         "(%-18s . %08d)\t; %c")
 749        ((>= (charset-dimension name) 2)
 750         "(%-18s . #x%04X)\t; %c")
 751        (t
 752         "(%-18s . #x%02X)\t; %c"))
 753       name
 754       (if (= (charset-iso-graphic-plane name) 1)
 755           (logior value
 756                   (cond ((= (charset-dimension name) 1)
 757                          #x80)
 758                         ((= (charset-dimension name) 2)
 759                          #x8080)
 760                         ((= (charset-dimension name) 3)
 761                          #x808080)
 762                         (t 0)))
 763         value)
 764       (char-db-decode-isolated-char name value)))
 765     (if (and (= (charset-chars name) 94)
 766              (= (charset-dimension name) 2))
 767         (insert (format " [%02d-%02d]"
 768                         (- (lsh value -8) 32)
 769                         (- (logand value 255) 32))))
 770     )
 771    (t
 772     (insert (format "(%-18s . %s)" name value))
 773     ))
 774   (insert line-breaking))
 775
 776 (defun char-db-insert-relation-feature (char name value line-breaking
 777                                              ccss readable)
 778   (insert (format "(%-18s%s " name line-breaking))
 779   (let ((lbs (concat "\n" (make-string (current-column) ?\ )))
 780         separator cell sources required-features
 781         ret)
 782     (while (consp value)
 783       (setq cell (car value))
 784       (if (integerp cell)
 785           (setq cell (decode-char '=ucs cell)))
 786       (cond
 787        ((eq name '->subsumptive)
 788         (when (or (not (some (lambda (atr)
 789                                (get-char-attribute cell atr))
 790                              char-db-ignored-attributes))
 791                   (some (lambda (ccs)
 792                           (encode-char cell ccs 'defined-only))
 793                         ccss))
 794           (if separator
 795               (insert lbs))
 796           (let ((char-db-ignored-attributes
 797                  (cons '<-subsumptive
 798                        char-db-ignored-attributes)))
 799             (insert-char-attributes cell readable nil nil 'for-sub-node))
 800           (setq separator lbs))
 801         )
 802        ((characterp cell)
 803         (setq sources
 804               (get-char-attribute
 805                char (intern (format "%s*sources" name))))
 806         (setq required-features nil)
 807         (dolist (source sources)
 808           (cond
 809            ((memq source '(JP
 810                            JP/Jouyou
 811                            shinjigen shinjigen@1ed shinjigen@rev))
 812             (setq required-features
 813                   (union required-features
 814                          '(=jis-x0208
 815                            =jis-x0208@1990
 816                            =jis-x0213-1@2000
 817                            =jis-x0213-1@2004
 818                            =jis-x0213-2
 819                            =jis-x0212
 820                            =jis-x0208@1983
 821                            =jis-x0208@1978
 822                            =shinjigen
 823                            =shinjigen@1ed
 824                            =shinjigen@rev
 825                            =shinjigen/+p@rev))))
 826            ((eq source 'CN)
 827             (setq required-features
 828                   (union required-features
 829                          '(=gb2312
 830                            =gb12345
 831                            =iso-ir165)))))
 832           (cond
 833            ((find-charset (setq ret (intern (format "=%s" source))))
 834             (setq required-features
 835                   (cons ret required-features)))
 836            (t (setq required-features
 837                     (cons source required-features)))))
 838         (cond ((string-match "@JP" (symbol-name name))
 839                (setq required-features
 840                      (union required-features
 841                             '(=jis-x0208
 842                               =jis-x0208@1990
 843                               =jis-x0213-1-2000
 844                               =jis-x0213-2-2000
 845                               =jis-x0212
 846                               =jis-x0208@1983
 847                               =jis-x0208@1978))))
 848               ((string-match "@CN" (symbol-name name))
 849                (setq required-features
 850                      (union required-features
 851                             '(=gb2312
 852                               =gb12345
 853                               =iso-ir165)))))
 854         (if separator
 855             (insert lbs))
 856         (if readable
 857             (insert (format "%S" cell))
 858           (char-db-insert-char-spec cell readable
 859                                     nil
 860                                     required-features))
 861         (setq separator lbs))
 862        ((consp cell)
 863         (if separator
 864             (insert lbs))
 865         (if (consp (car cell))
 866             (char-db-insert-char-spec cell readable)
 867           (char-db-insert-char-reference cell readable))
 868         (setq separator lbs))
 869        (t
 870         (if separator
 871             (insert separator))
 872         (insert (prin1-to-string cell))
 873         (setq separator " ")))
 874       (setq value (cdr value)))
 875     (insert ")")
 876     (insert line-breaking)))
 877
 878 (defun insert-char-attributes (char &optional readable attributes column
 879                                     for-sub-node)
 880   (unless column
 881     (setq column (current-column)))
 882   (let (name value ; has-long-ccs-name
 883         rest
 884         radical strokes
 885         (line-breaking
 886          (concat "\n" (make-string (1+ column) ?\ )))
 887         lbs cell separator ret
 888         key al cal
 889         dest-ccss ; sources required-features
 890         ccss)
 891     (let (atr-d)
 892       (setq attributes
 893             (sort (if attributes
 894                       (if (consp attributes)
 895                           (progn
 896                             (dolist (name attributes)
 897                               (unless (memq name char-db-ignored-attributes)
 898                                 (if (find-charset name)
 899                                     (push name ccss))
 900                                 (push name atr-d)))
 901                             atr-d))
 902                     (dolist (name (char-attribute-list))
 903                       (unless (memq name char-db-ignored-attributes)
 904                         (if (find-charset name)
 905                             (push name ccss))
 906                         (push name atr-d)))
 907                     atr-d)
 908                   #'char-attribute-name<)))
 909     (insert "(")
 910     (when (memq '<-subsumptive attributes)
 911       (when (or readable (not for-sub-node))
 912         (when (setq value (get-char-attribute char '<-subsumptive))
 913           (char-db-insert-relation-feature char '<-subsumptive value
 914                                            line-breaking
 915                                            ccss readable)))
 916       (setq attributes (delq '<-subsumptive attributes)))
 917     (when (and (memq '<-denotational attributes)
 918                (setq value (get-char-attribute char '<-denotational)))
 919       (char-db-insert-relation-feature char '<-denotational value
 920                                        line-breaking
 921                                        ccss readable)
 922       (setq attributes (delq '<-denotational attributes)))
 923     (when (and (memq '<-denotational@component attributes)
 924                (setq value (get-char-attribute char '<-denotational@component)))
 925       (char-db-insert-relation-feature char '<-denotational@component value
 926                                        line-breaking
 927                                        ccss readable)
 928       (setq attributes (delq '<-denotational@component attributes)))
 929     (when (and (memq 'name attributes)
 930                (setq value (get-char-attribute char 'name)))
 931       (insert (format
 932                (if (> (+ (current-column) (length value)) 48)
 933                    "(name . %S)%s"
 934                  "(name               . %S)%s")
 935                value line-breaking))
 936       (setq attributes (delq 'name attributes))
 937       )
 938     (when (and (memq 'name* attributes)
 939                (setq value (get-char-attribute char 'name*)))
 940       (insert (format
 941                (if (> (+ (current-column) (length value)) 48)
 942                    "(name* . %S)%s"
 943                  "(name*              . %S)%s")
 944                value line-breaking))
 945       (setq attributes (delq 'name* attributes))
 946       )
 947     (when (and (memq 'script attributes)
 948                (setq value (get-char-attribute char 'script)))
 949       (insert (format "(script\t\t%s)%s"
 950                       (mapconcat (function prin1-to-string)
 951                                  value " ")
 952                       line-breaking))
 953       (setq attributes (delq 'script attributes))
 954       )
 955     (dolist (name '(=>ucs =>ucs*))
 956       (when (and (memq name attributes)
 957                  (setq value (get-char-attribute char name)))
 958         (insert (format "(%-18s . #x%04X)\t; %c%s"
 959                         name value (decode-char '=ucs value)
 960                         line-breaking))
 961         (setq attributes (delq name attributes))))
 962     (dolist (name '(=>ucs@gb =>ucs@big5))
 963       (when (and (memq name attributes)
 964                  (setq value (get-char-attribute char name)))
 965         (insert (format "(%-18s . #x%04X)\t; %c%s"
 966                         name value
 967                         (decode-char (intern
 968                                       (concat "="
 969                                               (substring
 970                                                (symbol-name name) 2)))
 971                                      value)
 972                         line-breaking))
 973         (setq attributes (delq name attributes))
 974         ))
 975     ;; (dolist (name '(=>daikanwa))
 976     ;;   (when (and (memq name attributes)
 977     ;;              (setq value (get-char-attribute char name)))
 978     ;;     (insert
 979     ;;      (if (integerp value)
 980     ;;          (format "(%-18s . %05d)\t; %c%s"
 981     ;;                  name value (decode-char '=daikanwa value)
 982     ;;                  line-breaking)
 983     ;;        (format "(%-18s %s)\t; %c%s"
 984     ;;                name
 985     ;;                (mapconcat (function prin1-to-string)
 986     ;;                           value " ")
 987     ;;                (char-representative-of-daikanwa char)
 988     ;;                line-breaking)))
 989     ;;     (setq attributes (delq name attributes))))
 990     (when (and (memq 'general-category attributes)
 991                (setq value (get-char-attribute char 'general-category)))
 992       (insert (format
 993                "(general-category\t%s) ; %s%s"
 994                (mapconcat (lambda (cell)
 995                             (format "%S" cell))
 996                           value " ")
 997                (cond ((rassoc value unidata-normative-category-alist)
 998                       "Normative Category")
 999                      ((rassoc value unidata-informative-category-alist)
1000                       "Informative Category")
1001                      (t
1002                       "Unknown Category"))
1003                line-breaking))
1004       (setq attributes (delq 'general-category attributes))
1005       )
1006     (when (and (memq 'bidi-category attributes)
1007                (setq value (get-char-attribute char 'bidi-category)))
1008       (insert (format "(bidi-category\t. %S)%s"
1009                       value
1010                       line-breaking))
1011       (setq attributes (delq 'bidi-category attributes))
1012       )
1013     (unless (or (not (memq 'mirrored attributes))
1014                 (eq (setq value (get-char-attribute char 'mirrored 'empty))
1015                     'empty))
1016       (insert (format "(mirrored\t\t. %S)%s"
1017                       value
1018                       line-breaking))
1019       (setq attributes (delq 'mirrored attributes))
1020       )
1021     (cond
1022      ((and (memq 'decimal-digit-value attributes)
1023            (setq value (get-char-attribute char 'decimal-digit-value)))
1024       (insert (format "(decimal-digit-value . %S)%s"
1025                       value
1026                       line-breaking))
1027       (setq attributes (delq 'decimal-digit-value attributes))
1028       (when (and (memq 'digit-value attributes)
1029                  (setq value (get-char-attribute char 'digit-value)))
1030         (insert (format "(digit-value\t . %S)%s"
1031                         value
1032                         line-breaking))
1033         (setq attributes (delq 'digit-value attributes))
1034         )
1035       (when (and (memq 'numeric-value attributes)
1036                  (setq value (get-char-attribute char 'numeric-value)))
1037         (insert (format "(numeric-value\t . %S)%s"
1038                         value
1039                         line-breaking))
1040         (setq attributes (delq 'numeric-value attributes))
1041         )
1042       )
1043      (t
1044       (when (and (memq 'digit-value attributes)
1045                  (setq value (get-char-attribute char 'digit-value)))
1046         (insert (format "(digit-value\t. %S)%s"
1047                         value
1048                         line-breaking))
1049         (setq attributes (delq 'digit-value attributes))
1050         )
1051       (when (and (memq 'numeric-value attributes)
1052                  (setq value (get-char-attribute char 'numeric-value)))
1053         (insert (format "(numeric-value\t. %S)%s"
1054                         value
1055                         line-breaking))
1056         (setq attributes (delq 'numeric-value attributes))
1057         )))
1058     (when (and (memq 'iso-10646-comment attributes)
1059                (setq value (get-char-attribute char 'iso-10646-comment)))
1060       (insert (format "(iso-10646-comment\t. %S)%s"
1061                       value
1062                       line-breaking))
1063       (setq attributes (delq 'iso-10646-comment attributes))
1064       )
1065     (when (and (memq 'morohashi-daikanwa attributes)
1066                (setq value (get-char-attribute char 'morohashi-daikanwa)))
1067       (insert (format "(morohashi-daikanwa\t%s)%s"
1068                       (mapconcat (function prin1-to-string) value " ")
1069                       line-breaking))
1070       (setq attributes (delq 'morohashi-daikanwa attributes))
1071       )
1072     (setq radical nil
1073           strokes nil)
1074     (when (and (memq 'ideographic-radical attributes)
1075                (setq value (get-char-attribute char 'ideographic-radical)))
1076       (setq radical value)
1077       (insert (format "(ideographic-radical . %S)\t; %c%s"
1078                       radical
1079                       (ideographic-radical radical)
1080                       line-breaking))
1081       (setq attributes (delq 'ideographic-radical attributes))
1082       )
1083     (when (and (memq 'shuowen-radical attributes)
1084                (setq value (get-char-attribute char 'shuowen-radical)))
1085       (insert (format "(shuowen-radical\t. %S)\t; %c%s"
1086                       value
1087                       (shuowen-radical value)
1088                       line-breaking))
1089       (setq attributes (delq 'shuowen-radical attributes))
1090       )
1091     (let (key)
1092       (dolist (domain
1093                (append
1094                 char-db-feature-domains
1095                 (let (dest domain)
1096                   (dolist (feature (char-attribute-list))
1097                     (setq feature (symbol-name feature))
1098                     (when (string-match
1099                            "\\(radical\\|strokes\\)@\\([^@*]+\\)\\(\\*\\|$\\)"
1100                            feature)
1101                       (setq domain (intern (match-string 2 feature)))
1102                      (unless (memq domain dest)
1103                        (setq dest (cons domain dest)))))
1104                   (sort dest #'string<))))
1105         (setq key (intern (format "%s@%s" 'ideographic-radical domain)))
1106         (when (and (memq key attributes)
1107                    (setq value (get-char-attribute char key)))
1108           (setq radical value)
1109           (insert (format "(%s . %S)\t; %c%s"
1110                           key
1111                           radical
1112                           (ideographic-radical radical)
1113                           line-breaking))
1114           (setq attributes (delq key attributes))
1115           )
1116         (setq key (intern (format "%s@%s" 'ideographic-strokes domain)))
1117         (when (and (memq key attributes)
1118                    (setq value (get-char-attribute char key)))
1119           (setq strokes value)
1120           (insert (format "(%s . %S)%s"
1121                           key
1122                           strokes
1123                           line-breaking))
1124           (setq attributes (delq key attributes))
1125           )
1126         (setq key (intern (format "%s@%s" 'total-strokes domain)))
1127         (when (and (memq key attributes)
1128                    (setq value (get-char-attribute char key)))
1129           (insert (format "(%s       . %S)%s"
1130                           key
1131                           value
1132                           line-breaking))
1133           (setq attributes (delq key attributes))
1134           )
1135         (dolist (feature '(ideographic-radical
1136                            ideographic-strokes
1137                            total-strokes))
1138           (setq key (intern (format "%s@%s*sources" feature domain)))
1139           (when (and (memq key attributes)
1140                      (setq value (get-char-attribute char key)))
1141             (insert (format "(%s%s" key line-breaking))
1142             (dolist (cell value)
1143               (insert (format " %s" cell)))
1144             (insert ")")
1145             (insert line-breaking)
1146             (setq attributes (delq key attributes))
1147             ))
1148         ))
1149     (when (and (memq 'ideographic-strokes attributes)
1150                (setq value (get-char-attribute char 'ideographic-strokes)))
1151       (setq strokes value)
1152       (insert (format "(ideographic-strokes . %S)%s"
1153                       strokes
1154                       line-breaking))
1155       (setq attributes (delq 'ideographic-strokes attributes))
1156       )
1157     (when (and (memq 'kangxi-radical attributes)
1158                (setq value (get-char-attribute char 'kangxi-radical)))
1159       (unless (eq value radical)
1160         (insert (format "(kangxi-radical\t . %S)\t; %c%s"
1161                         value
1162                         (ideographic-radical value)
1163                         line-breaking))
1164         (or radical
1165             (setq radical value)))
1166       (setq attributes (delq 'kangxi-radical attributes))
1167       )
1168     (when (and (memq 'kangxi-strokes attributes)
1169                (setq value (get-char-attribute char 'kangxi-strokes)))
1170       (unless (eq value strokes)
1171         (insert (format "(kangxi-strokes\t . %S)%s"
1172                         value
1173                         line-breaking))
1174         (or strokes
1175             (setq strokes value)))
1176       (setq attributes (delq 'kangxi-strokes attributes))
1177       )
1178     ;; (when (and (memq 'japanese-radical attributes)
1179     ;;            (setq value (get-char-attribute char 'japanese-radical)))
1180     ;;   (unless (eq value radical)
1181     ;;     (insert (format "(japanese-radical\t . %S)\t; %c%s"
1182     ;;                     value
1183     ;;                     (ideographic-radical value)
1184     ;;                     line-breaking))
1185     ;;     (or radical
1186     ;;         (setq radical value)))
1187     ;;   (setq attributes (delq 'japanese-radical attributes))
1188     ;;   )
1189     (when (and (memq 'japanese-strokes attributes)
1190                (setq value (get-char-attribute char 'japanese-strokes)))
1191       (unless (eq value strokes)
1192         (insert (format "(japanese-strokes\t . %S)%s"
1193                         value
1194                         line-breaking))
1195         (or strokes
1196             (setq strokes value)))
1197       (setq attributes (delq 'japanese-strokes attributes))
1198       )
1199     (when (and (memq 'cns-radical attributes)
1200                (setq value (get-char-attribute char 'cns-radical)))
1201       (insert (format "(cns-radical\t . %S)\t; %c%s"
1202                       value
1203                       (ideographic-radical value)
1204                       line-breaking))
1205       (setq attributes (delq 'cns-radical attributes))
1206       )
1207     (when (and (memq 'cns-strokes attributes)
1208                (setq value (get-char-attribute char 'cns-strokes)))
1209       (unless (eq value strokes)
1210         (insert (format "(cns-strokes\t . %S)%s"
1211                         value
1212                         line-breaking))
1213         (or strokes
1214             (setq strokes value)))
1215       (setq attributes (delq 'cns-strokes attributes))
1216       )
1217     ;; (when (and (memq 'shinjigen-1-radical attributes)
1218     ;;            (setq value (get-char-attribute char 'shinjigen-1-radical)))
1219     ;;   (unless (eq value radical)
1220     ;;     (insert (format "(shinjigen-1-radical . %S)\t; %c%s"
1221     ;;                     value
1222     ;;                     (ideographic-radical value)
1223     ;;                     line-breaking))
1224     ;;     (or radical
1225     ;;         (setq radical value)))
1226     ;;   (setq attributes (delq 'shinjigen-1-radical attributes))
1227     ;;   )
1228     ;; (when (and (memq 'ideographic- attributes)
1229     ;;            (setq value (get-char-attribute char 'ideographic-)))
1230     ;;   (insert "(ideographic-       ")
1231     ;;   (setq lbs (concat "\n" (make-string (current-column) ?\ ))
1232     ;;         separator nil)
1233     ;;   (while (consp value)
1234     ;;     (setq cell (car value))
1235     ;;     (if (integerp cell)
1236     ;;         (setq cell (decode-char '=ucs cell)))
1237     ;;     (cond ((characterp cell)
1238     ;;            (if separator
1239     ;;                (insert lbs))
1240     ;;            (if readable
1241     ;;                (insert (format "%S" cell))
1242     ;;              (char-db-insert-char-spec cell readable))
1243     ;;            (setq separator lbs))
1244     ;;           ((consp cell)
1245     ;;            (if separator
1246     ;;                (insert lbs))
1247     ;;            (if (consp (car cell))
1248     ;;                (char-db-insert-char-spec cell readable)
1249     ;;              (char-db-insert-char-reference cell readable))
1250     ;;            (setq separator lbs))
1251     ;;           (t
1252     ;;            (if separator
1253     ;;                (insert separator))
1254     ;;            (insert (prin1-to-string cell))
1255     ;;            (setq separator " ")))
1256     ;;     (setq value (cdr value)))
1257     ;;   (insert ")")
1258     ;;   (insert line-breaking)
1259     ;;   (setq attributes (delq 'ideographic- attributes)))
1260     (when (and (memq 'total-strokes attributes)
1261                (setq value (get-char-attribute char 'total-strokes)))
1262       (insert (format "(total-strokes       . %S)%s"
1263                       value
1264                       line-breaking))
1265       (setq attributes (delq 'total-strokes attributes))
1266       )
1267     ;; (when (and (memq '->ideograph attributes)
1268     ;;            (setq value (get-char-attribute char '->ideograph)))
1269     ;;   (insert (format "(->ideograph\t%s)%s"
1270     ;;                   (mapconcat (lambda (code)
1271     ;;                                (cond ((symbolp code)
1272     ;;                                       (symbol-name code))
1273     ;;                                      ((integerp code)
1274     ;;                                       (format "#x%04X" code))
1275     ;;                                      (t
1276     ;;                                       (format "%s %S"
1277     ;;                                               line-breaking code))))
1278     ;;                              value " ")
1279     ;;                   line-breaking))
1280     ;;   (setq attributes (delq '->ideograph attributes))
1281     ;;   )
1282     ;; (when (and (memq '->decomposition attributes)
1283     ;;            (setq value (get-char-attribute char '->decomposition)))
1284     ;;   (insert (format "(->decomposition\t%s)%s"
1285     ;;                   (mapconcat (lambda (code)
1286     ;;                                (cond ((symbolp code)
1287     ;;                                       (symbol-name code))
1288     ;;                                      ((characterp code)
1289     ;;                                       (if readable
1290     ;;                                           (format "%S" code)
1291     ;;                                         (format "#x%04X"
1292     ;;                                                 (char-int code))
1293     ;;                                         ))
1294     ;;                                      ((integerp code)
1295     ;;                                       (format "#x%04X" code))
1296     ;;                                      (t
1297     ;;                                       (format "%s%S" line-breaking code))))
1298     ;;                              value " ")
1299     ;;                   line-breaking))
1300     ;;   (setq attributes (delq '->decomposition attributes))
1301     ;;   )
1302     (if (equal (get-char-attribute char '->titlecase)
1303                (get-char-attribute char '->uppercase))
1304         (setq attributes (delq '->titlecase attributes)))
1305     (when (and (memq '->mojikyo attributes)
1306                (setq value (get-char-attribute char '->mojikyo)))
1307       (insert (format "(->mojikyo\t\t. %06d)\t; %c%s"
1308                       value (decode-char 'mojikyo value)
1309                       line-breaking))
1310       (setq attributes (delq '->mojikyo attributes))
1311       )
1312     ;; (when (and (memq 'hanyu-dazidian-vol attributes)
1313     ;;            (setq value (get-char-attribute char 'hanyu-dazidian-vol)))
1314     ;;   (insert (format "(hanyu-dazidian-vol  . %d)%s"
1315     ;;                   value line-breaking))
1316     ;;   (setq attributes (delq 'hanyu-dazidian-vol attributes))
1317     ;;   )
1318     ;; (when (and (memq 'hanyu-dazidian-page attributes)
1319     ;;            (setq value (get-char-attribute char 'hanyu-dazidian-page)))
1320     ;;   (insert (format "(hanyu-dazidian-page . %d)%s"
1321     ;;                   value line-breaking))
1322     ;;   (setq attributes (delq 'hanyu-dazidian-page attributes))
1323     ;;   )
1324     ;; (when (and (memq 'hanyu-dazidian-char attributes)
1325     ;;            (setq value (get-char-attribute char 'hanyu-dazidian-char)))
1326     ;;   (insert (format "(hanyu-dazidian-char . %d)%s"
1327     ;;                   value line-breaking))
1328     ;;   (setq attributes (delq 'hanyu-dazidian-char attributes))
1329     ;;   )
1330     (unless readable
1331       (dolist (ignored '(composition
1332                          ->denotational <-subsumptive ->ucs-unified
1333                          ->ideographic-component-forms))
1334         (setq attributes (delq ignored attributes))))
1335     (while attributes
1336       (setq name (car attributes))
1337       (unless (eq (setq value (get-char-attribute char name 'value-is-empty))
1338                   'value-is-empty)
1339         (cond ((setq ret (find-charset name))
1340                (setq name (charset-name ret))
1341                (when (not (memq name dest-ccss))
1342                  (setq dest-ccss (cons name dest-ccss))
1343                  (char-db-insert-ccs-feature name value line-breaking))
1344                )
1345               ((string-match "^=>ucs@" (symbol-name name))
1346                (insert (format "(%-18s . #x%04X)\t; %c%s"
1347                                name value (decode-char '=ucs value)
1348                                line-breaking))
1349                )
1350               ((eq name 'jisx0208-1978/4X)
1351                (insert (format "(%-18s . #x%04X)%s"
1352                                name value
1353                                line-breaking))
1354                )
1355               ((and
1356                 (not readable)
1357                 (not (eq name '->subsumptive))
1358                 (not (eq name '->uppercase))
1359                 (not (eq name '->lowercase))
1360                 (not (eq name '->titlecase))
1361                 (not (eq name '->canonical))
1362                 (not (eq name '->Bopomofo))
1363                 (not (eq name '->mistakable))
1364                 (not (eq name '->ideographic-variants))
1365                 (null (get-char-attribute
1366                        char (intern (format "%s*sources" name))))
1367                 (not (string-match "\\*sources$" (symbol-name name)))
1368                 (null (get-char-attribute
1369                        char (intern (format "%s*note" name))))
1370                 (not (string-match "\\*note$" (symbol-name name)))
1371                 (or (eq name '<-identical)
1372                     (eq name '<-uppercase)
1373                     (eq name '<-lowercase)
1374                     (eq name '<-titlecase)
1375                     (eq name '<-canonical)
1376                     (eq name '<-ideographic-variants)
1377                     ;; (eq name '<-synonyms)
1378                     (string-match "^<-synonyms" (symbol-name name))
1379                     (eq name '<-mistakable)
1380                     (when (string-match "^->" (symbol-name name))
1381                       (cond
1382                        ((string-match "^->fullwidth" (symbol-name name))
1383                         (not (and (consp value)
1384                                   (characterp (car value))
1385                                   (encode-char
1386                                    (car value) '=ucs 'defined-only)))
1387                         )
1388                        (t)))
1389                     ))
1390                )
1391               ((or (eq name 'ideographic-structure)
1392                    (eq name 'ideographic-combination)
1393                    ;; (eq name 'ideographic-)
1394                    (eq name '=decomposition)
1395                    (char-feature-base-name= '=decomposition name)
1396                    (char-feature-base-name= '=>decomposition name)
1397                    ;; (string-match "^=>*decomposition\\(@[^*]+\\)?$"
1398                    ;;               (symbol-name name))
1399                    (string-match "^\\(->\\|<-\\)[^*]*$" (symbol-name name))
1400                    (string-match "^\\(->\\|<-\\)[^*]*\\*sources$"
1401                                  (symbol-name name))
1402                    )
1403                (char-db-insert-relation-feature char name value
1404                                                 line-breaking
1405                                                 ccss readable))
1406               ;; ((memq name '(ideograph=
1407               ;;               original-ideograph-of
1408               ;;               ancient-ideograph-of
1409               ;;               vulgar-ideograph-of
1410               ;;               wrong-ideograph-of
1411               ;;               ;; simplified-ideograph-of
1412               ;;               ideographic-variants
1413               ;;               ;; ideographic-different-form-of
1414               ;;               ))
1415               ;;  (insert (format "(%-18s%s " name line-breaking))
1416               ;;  (setq lbs (concat "\n" (make-string (current-column) ?\ ))
1417               ;;        separator nil)
1418               ;;  (while (consp value)
1419               ;;    (setq cell (car value))
1420               ;;    (if (and (consp cell)
1421               ;;             (consp (car cell)))
1422               ;;        (progn
1423               ;;          (if separator
1424               ;;              (insert lbs))
1425               ;;          (char-db-insert-alist cell readable)
1426               ;;          (setq separator lbs))
1427               ;;      (if separator
1428               ;;          (insert separator))
1429               ;;      (insert (prin1-to-string cell))
1430               ;;      (setq separator " "))
1431               ;;    (setq value (cdr value)))
1432               ;;  (insert ")")
1433               ;;  (insert line-breaking))
1434               ((consp value)
1435                (insert (format "(%-18s " name))
1436                (setq lbs (concat "\n" (make-string (current-column) ?\ ))
1437                      separator nil)
1438                (while (consp value)
1439                  (setq cell (car value))
1440                  (if (and (consp cell)
1441                           (consp (car cell))
1442                           (setq ret (condition-case nil
1443                                         (find-char cell)
1444                                       (error nil))))
1445                      (progn
1446                        (setq rest cell
1447                              al nil
1448                              cal nil)
1449                        (while rest
1450                          (setq key (car (car rest)))
1451                          (if (find-charset key)
1452                              (setq cal (cons key cal))
1453                            (setq al (cons key al)))
1454                          (setq rest (cdr rest)))
1455                        (if separator
1456                            (insert lbs))
1457                        (insert-char-attributes ret
1458                                                readable
1459                                                al ; cal
1460                                                nil 'for-sub-node)
1461                        (setq separator lbs))
1462                    (setq ret (prin1-to-string cell))
1463                    (if separator
1464                        (if (< (+ (current-column)
1465                                  (length ret)
1466                                  (length separator))
1467                               76)
1468                            (insert separator)
1469                          (insert lbs)))
1470                    (insert ret)
1471                    (setq separator " "))
1472                  (setq value (cdr value)))
1473                (insert ")")
1474                (insert line-breaking))
1475               (t
1476                (insert (format "(%-18s" name))
1477                (setq ret (prin1-to-string value))
1478                (unless (< (+ (current-column)
1479                              (length ret)
1480                              3)
1481                           76)
1482                  (insert line-breaking))
1483                (insert " . " ret ")" line-breaking)
1484                ;; (insert (format "(%-18s . %S)%s"
1485                ;;                 name value
1486                ;;                 line-breaking))
1487                )
1488               ))
1489       (setq attributes (cdr attributes)))
1490     (insert ")")))
1491
1492 (defun insert-char-data (char &optional readable
1493                               attributes)
1494   (save-restriction
1495     (narrow-to-region (point)(point))
1496     (insert "(define-char
1497   '")
1498     (insert-char-attributes char readable attributes)
1499     (insert ")\n")
1500     (goto-char (point-min))
1501     (while (re-search-forward "[ \t]+$" nil t)
1502       (replace-match ""))
1503     ;; from tabify.
1504     (goto-char (point-min))
1505     (while (re-search-forward "[ \t][ \t][ \t]*" nil t)
1506       (let ((column (current-column))
1507             (indent-tabs-mode t))
1508         (delete-region (match-beginning 0) (point))
1509         (indent-to column)))
1510     (goto-char (point-max))
1511     ;; (tabify (point-min)(point-max))
1512     ))
1513
1514 (defun insert-char-data-with-variant (char &optional printable
1515                                            no-ucs-unified
1516                                            script excluded-script)
1517   (insert-char-data char printable)
1518   (let ((variants (char-variants char))
1519         rest
1520         variant vs ret)
1521     (setq variants (sort variants #'<))
1522     (setq rest variants)
1523     (setq variants (cons char variants))
1524     (while rest
1525       (setq variant (car rest))
1526       (unless (get-char-attribute variant '<-subsumptive)
1527         (if (and (or (null script)
1528                      (null (setq vs (get-char-attribute variant 'script)))
1529                      (memq script vs))
1530                  (or (null excluded-script)
1531                      (null (setq vs (get-char-attribute variant 'script)))
1532                      (not (memq excluded-script vs))))
1533             (unless (and no-ucs-unified (get-char-attribute variant '=ucs))
1534               (insert-char-data variant printable)
1535               (if (setq ret (char-variants variant))
1536                   (while ret
1537                     (or (memq (car ret) variants)
1538                         ;; (get-char-attribute (car ret) '<-subsumptive)
1539                         (setq rest (nconc rest (list (car ret)))))
1540                     (setq ret (cdr ret)))))))
1541       (setq rest (cdr rest)))))
1542
1543 (defun insert-char-range-data (min max &optional script excluded-script)
1544   (let ((code min)
1545         char)
1546     (while (<= code max)
1547       (setq char (decode-char '=ucs code))
1548       (if (encode-char char '=ucs 'defined-only)
1549           (insert-char-data-with-variant char nil 'no-ucs-unified
1550                                          script excluded-script))
1551       (setq code (1+ code)))))
1552
1553 (defun write-char-range-data-to-file (min max file
1554                                           &optional script excluded-script)
1555   (let ((coding-system-for-write char-db-file-coding-system))
1556     (with-temp-buffer
1557       (insert (format ";; -*- coding: %s -*-\n"
1558                       char-db-file-coding-system))
1559       (insert-char-range-data min max script excluded-script)
1560       (write-region (point-min)(point-max) file))))
1561
1562 (defvar what-character-original-window-configuration)
1563
1564 ;;;###autoload
1565 (defun what-char-definition (char)
1566   (interactive (list (char-after)))
1567   (let ((buf (get-buffer-create "*Character Description*"))
1568         (the-buf (current-buffer))
1569         (win-conf (current-window-configuration)))
1570     (pop-to-buffer buf)
1571     (make-local-variable 'what-character-original-window-configuration)
1572     (setq what-character-original-window-configuration win-conf)
1573     (setq buffer-read-only nil)
1574     (erase-buffer)
1575     (condition-case err
1576         (progn
1577           (insert-char-data-with-variant char 'printable)
1578           (unless (char-attribute-alist char)
1579             (insert (format ";; = %c\n"
1580                             (let* ((rest (split-char char))
1581                                    (ccs (pop rest))
1582                                    (code (pop rest)))
1583                               (while rest
1584                                 (setq code (logior (lsh code 8)
1585                                                    (pop rest))))
1586                               (decode-char ccs code)))))
1587           ;; (char-db-update-comment)
1588           (set-buffer-modified-p nil)
1589           (view-mode the-buf (lambda (buf)
1590                                (set-window-configuration
1591                                 what-character-original-window-configuration)
1592                                ))
1593           (goto-char (point-min)))
1594       (error (progn
1595                (set-window-configuration
1596                 what-character-original-window-configuration)
1597                (signal (car err) (cdr err)))))))
1598
1599
1600 ;;; @ end
1601 ;;;
1602
1603 (provide 'char-db-util)
1604
1605 ;;; char-db-util.el ends here